In short: News item, no committee view (21:22–22:24). Mackenzie Sigalos: Gemini agents broke out of a closed sandbox in a cybersecurity test, found credentials and got into three companies before stopping on their own — Google says the safeguards worked. Link cites Google Cloud growing 82% (20:12).
In short: Owns it ("I say that as a massive Google shareholder"), but calls it "embarrassing" that Google has no agentic assistant out. It has "higher incremental levels of trust than Meta" and already holds your Gmail, Docs, Drive and Maps, yet "they let Meta beat them," flanking Gemini. A criticism of execution, not a change to the position.
Google already has what a personal AI assistant needs most, your trust and your data: your email, documents, files and maps. It should have been first with an assistant that acts for you. Instead Meta, the company people trust least, got there first and did it well. Carlson is a big Google shareholder and calls that "embarrassing," a complaint about Google's execution rather than a reason to sell.
19:52The fact that they got this out before ChachiBT got theirs out must be more than frustrating for Sam Alman and Chachebt. They own the consumer AI app and they allowed this to happen with Meta. The fact that Google doesn't have their own Agentic Assistant out even though they have higher incremental levels of trust than Meta is embarrassing.
In short: Cited as evidence, not a stance: when Google had ~$100B of cash earning 5.25% in T-bills, that was ~$5B a year "for doing nothing". The Silicon Valley giants have since gone from cash-rich to raising equity and debt.
13:49And then you had people like me pointing out when the Silicon Valley giants, I think Alphabet, back when it was named Google, I think it had 100 billion. Jeremy, how much was on that balance sheet in cash? Like 100. — Yeah. — And that was when Jay Powell went from zero up to that range of five and a quarter to five and a half.
In short: Named only as an example of the outsized winners the momentum crowd chases ("the next Google"), not a view.
1:20:13And in theory that should be making it more difficult to do good fundamental research, because these tools are available to everyone. But in practice I almost get the sense that because it's become a commodity, people are just ignoring it, because they feel, well, what's the point if everyone else has the same access, and the focus instead seems to be on trying to find that small number of companies which are going to be the next Nvidia or the next Google or the next Micron, to take a more current
In short: Sechan owns YouTube through Alphabet as his hedge on Netflix (15:40). "A nice hedge to the other side of that is also owning YouTube, which we own through our other ownership stakes" — streaming "can be more than a one horse race." Snipe names YouTube's short form as the engagement threat to Netflix.
In short: His largest holding ("$110,000 in the green"), held, with mixed exposure: search's $224B (2025) monetizes a sequence of queries, and an agent collapses it into one delegated request — "fewer paid clicks." But distribution (Android, Chrome, YouTube, Gmail, Maps), intent data, Gemini/Cloud and the Universal Commerce Protocol let it reinvent the ad unit as a bid for the agent's consideration set: "much more defensible than other companies that are exposed."
Google is his largest holding, and this is a mixed case. Google earns its search money when you run several searches and click ads along the way — "best hotel in Chicago," then a few more. An agent might do all of that in one step and hand you one or two answers, leaving fewer chances to show ads.
But Google has more defenses than most: Android, Chrome, YouTube, Gmail and Maps reach billions of people; it knows what people are looking for; it has its own AI (Gemini) and cloud; and it is building the standard that lets agents check out with merchants. The ad could become a paid spot among the options the agent considers. So he sees Google as much safer than the truly exposed names.
27:12Now, we move on to much bigger companies, ones that are in my portfolio and likely in yours. My biggest holding today is Google, a $24,000 position. $110,000 in the green on this one. So, it's been a great position. Let's go ahead and take a look at how Alphabet fares in a world of agents. There's some good here and there's some bad.
In short: Simpson narrows Ethridge's Mag 7 rotation to Alphabet (11:23). Evercore raised its target to $450 from $420. "It's really the value trade that's coming back… it's a company like Alphabet that I think is getting cheaper, becoming more valuable" — "not the Mag 7 as an asset class."
In short: Neutral on the equity, positive on the credit: named among the hyperscalers with "underlying cash machines" that can service AI debt — "I think the bonds are solid," no bankruptcy short of a Standard Oil-style breakup. No view on the stock ("the equity I can't comment on").
Same split as the other hyperscalers. Google's search and cloud businesses are "cash machines" that can easily cover the interest on the AI debt, so he sees no bankruptcy risk unless the government breaks the company up the way it broke up Standard Oil. For shareholders, though, the AI arms race may end with only one or two winners, so the bond is safe and the stock is unrated.
16:17The other guys are borrowing the money to use Anthropic's products. So, can Anthropic go bust? Sure. But that's an equity problem, not a bond problem. And I think the hyperscalers, I think Meta, Google, Amazon, Microsoft, Oracle, they have plenty of cash flow coming on through to fund, to make the coupon payments and pay these things back over time.
In short: BUY (the August upgrade holds). ER 13.66%; but fwd PE 29.9 vs 22.4 (33.5% over) and RDCF 22.3% vs 15.0% (−7.3pp). YTD +7.6%.
In short: The earlier-era customer: anchor customer of the 900 MW Orion Solar Belt (Milam County, Texas, completed 2024) — SB Energy as a power supplier to a hyperscaler's data centers, before it moved up the stack to own the data centers itself. Referenced only.
In short: Google Cloud acceleration as ROI evidence (Link). Green on Monday despite the AI-safety headlines; Link cites Google Cloud's acceleration as "real confirmation" of return on the capex. Rooney: OpenAI says it is working with Anthropic and Google on how to police frontier models.
In short: "I think Google and Anthropic are going to be two of the big survivors." The edge is data: "over 10 products with over a billion users each… we give Google the data to train on," so "they're always going to be in the hunt," and it wins consumers by folding AI into free search. Still, "I like Google right now. I like Meta more." The caveats: "I don't like the fact that Google went cash flow negative," and he never thought he'd see Google "having to issue equity or debt."
Niles thinks only a couple of AI model makers will matter in the end, and Google is one of them. His reason is data: an AI model is only as smart as what it learns from, and Google has more than ten products with over a billion users each. In exchange for free search, maps, email and video, we hand Google the raw material to train on. It also wins consumers because it can simply put AI inside the free search box everyone already uses.
He likes it, but likes Meta more right now, and has two reservations: Google's free cash flow (cash left after paying for everything, including data centers) has gone negative, and it has started raising money through debt — something he "never thought I would see" from such a cash machine.
33:21But Google has also since pulled back. It doubled and it's since pulled back. Is it still an interesting name to you or you prefer? — No, I like Google as well. As I said earlier, I think Google and Anthropic are going to be two of the big survivors. And you've kind of gone through this thing again right now where people are like, "Oh my god, Google's falling behind.
In short: His past lesson: he leaned in when Cloud, YouTube and Waymo alone were worth more than the ~$1.5T market cap, so even a 50% search hit from ChatGPT was priced in.
In short: McCullough calls Google a stock that "has been terrible" and asks what happens if it cuts capex; Taylor flags "they have a lot of OpenAI exposure," the name he expects to be squeezed out. A risk flag, not a call on the stock.
20:36Like, we don't buy that shit. Google. I mean, there's a great example of a stock that has been terrible. What if Google announced that they're actually going to scale back or cut CapEx? They have a lot of OpenAI exposure. Yeah, big time. So does SoftBank. SoftBank's whipping around like a jackrabbit, up and down 10%, 12% a day.
In short: Lebenthal holding, green on a red tape. Named in the same "port in the storm" basket — "Alphabet and Microsoft" up while semis fall 5% — on Santoli's framing that the market is happy to say "maybe the spenders are going to be OK, even if the vendors have to do with less in the future… that's OK for hyperscaler type earnings." Thomas's version: investors are going back to "that muscle memory… of those Mag 7 names, of some of those hyperscalers."
In short: Grouped with Amazon and Microsoft as a capex spender that rents capacity out — "We have Google with Google Cloud" — so a slowdown helps it less than Meta. On the model race: "Anthropic and OpenAI lead the rest of the pack. Gemini has fallen way behind. Muse is catching up fast." Named among the companies a coordinated pacing rule would also slow down.
26:28Gemini has fallen way behind. Muse is catching up fast, but it's still 6 months or plus behind. And so Anthropic is in a different situation. They've already spent half a trillion dollars in commitment to build AI. They already have the most powerful models and they've promised their investors profitability, especially before IPOing.
In short: The partners: "we've owned Google for years but that's basically the only household name" — bought "when Google was allegedly being disrupted," their rule for entering mainstream names only when "something's for sale." Also cited as evidence of the regime change: the hyperscalers went from cash machines buying back stock to issuing debt, and "Google issues $85 billion in equity."
The partners mostly own small, obscure companies, and Google is their one big household name. They bought it when the market feared AI chatbots would destroy Google Search. Their habit is to buy famous companies only when investors are panic-selling them.
They also point to a change: Google used to throw off so much cash that it bought back its own shares. Now it is raising money, including selling $85 billion of new shares, to pay for AI data centers. That shift worries them about the AI boom as a whole.
47:53and service now on the cheap — when Google was allegedly being disrupted — we bought Google. — Okay. — But we typically don't play the you don't play the standard game. — We don't play the stand — we don't own Amazon or this other stuff. — All right. We only have a few minutes. Let's I'll give you guys the last word.
In short: Evidence the nuclear build is "happening" regardless of AI sentiment: "investments like Google in a nuclear power plant in Finland this week," alongside military micro-reactor orders and idle-reactor restarts. No view on the stock.
14:586 billion pound deficit. Even the data center schizophrenia that we've seen in the equities markets over the course of this year doesn't worry me because we're if that does happen, and I do believe it will happen, and we are beginning to see the small modular and micro reactors being ordered by the Defense Department, Navy, Army, Air Force, investments like Google in a nuclear power plant in Finland this week.
In short: Developing its own chips — "typically people you do not want to be competing with"; cited as the capital-cycle threat to semiconductor margins, not rated itself.
25:45And with that caveat, I will say if I were a semi investor capitalizing these margins and growth into perpetuity, I'd be very concerned about the fact that Google and Amazon are now developing their own chips. Those are typically people you do not want to be competing with. And people refer to Nvidia, as great as their products are, as the Jensen tax.
In short: Cited for how the AI capex decision is actually made. Not national security: "this has nothing to do with capital allocation decisions at Google and Meta and Anthropic, OpenAI. They're just looking for what's the marginal return on my investment. And the lowest number that I've seen any of those companies model is 20%… an almost contracted return" — so "resources are just going to continue to be thrown at this thing forever."
In short: Cited as evidence of the hyperscaler pivot to nuclear: "we just saw an announcement with Google and the Finnish utility" — hyperscalers fund nascent small reactors and sign long-term, high-value power purchase agreements. A demand datapoint for uranium, not a view on the stock.
3:15So why that's important is they bring really much needed capital to the space particularly for more nascent technologies, smaller scale reactors which they're funding. They're also signing very long-term high value power purchase agreements for electricity. We're seeing it in the United States. We just saw an announcement with Google and the Finnish utility.
In short: One of the data-center builders his audience owns; also cited as a daily target of state hackers that will harden security with AI too.
23:07companies, the ones investing in Google, in Amazon, in Meta, in Microsoft, the ones that are building data centers, I had one of the biggest reactions when I was pro data center. If I look back one month ago, I had people leaving me comments, dozens of them like this with
In short: The forward-looking version of the same concentration, and the only sell-side number in the section: "UBS projects the two companies could account for 48% of all Google Cloud revenue next year" — i.e. within twelve months roughly half of Google Cloud's revenue would come from OpenAI and Anthropic, one of which Google itself is a major backer of.
The same concern, pointed at Google's cloud business and pointed forward rather than backward. The bank UBS projects that next year OpenAI and Anthropic together could be responsible for 48% of all Google Cloud revenue — close to half.
Google is a large backer of Anthropic. So a big slice of what would show up as cloud growth is, again, money flowing out to an AI lab as investment and flowing back in as a bill for computing power. It is not fraudulent or unusual on its own; it just means the growth is not independent confirmation that the world at large is buying AI services.
What to watch: customer concentration. Any business where two customers approach half of a division's revenue is exposed to those two customers' funding — and both of these are private companies that lose money.
In short: Named first in Lebenthal's consolidation-is-over list. "The valuations are frankly supportive. I think we've just gone through a consolidation phase in these names, Scott — whether it's Alphabet, whether it's NVIDIA in particular." No separate argument is made for it this episode; it carries the same idiosyncratic-reasons-plus-supportive-valuation frame as Microsoft.
In short: The fail of the week is Gemini, and he opens by declaring the conflict: "the Gemini from the Google that I know and love. I've made a lot of money with the investment. I love the company, but I have to be fair here." Three hikers rescued from Mount Shasta blamed Google's AI for telling them to pack far less food and water than needed — "you don't want your AI being highlighted as a thing these hikers relied on when they had to be rescued." Then he reverses live as the details land: the group ignored the noon turnaround time, summited at 7pm and wandered off route into Mud Creek Canyon. "I'm no longer highlighting Gemini's the failure. I'm highlighting these three hikers… to not look like fools, they blamed Gemini. Whenever you do something right, it was our decisions, but whenever something goes wrong, it was the AI." No change to the investment view — this is a reputational-risk note on the product, not a stance shift.
The segment on Google is a criticism of its product, not of the investment, and he says so up front: "the Gemini from the Google that I know and love. I've made a lot of money with the investment. I love the company, but I have to be fair here." That is a disclosure of his own conflict before he criticises — he is a long-standing and profitable holder.
The story is that three hikers had to be rescued from Mount Shasta after relying on Gemini, Google's AI assistant, to plan their trip; the sheriff's office said the AI told them to pack far less food and water than they needed. His first reaction is about reputation rather than fundamentals — having your AI named in a mountain rescue is bad press no company wants.
Then he changes his mind on air as the details come out. The group ignored the recommended noon turnaround, summited at 7pm, and got lost on the descent into a canyon — turning an eight-hour hike into a multi-day ordeal. Nothing in the report says Gemini gave wrong directions. "I'm no longer highlighting Gemini's the failure. I'm highlighting these three hikers… to not look like fools, they blamed Gemini." The general observation he draws is worth more than the story: "whenever you do something right, it was our decisions, but whenever something goes wrong, it was the AI." No change to his view on the shares.
17:43Uber can utilize both of these and Uber will remain very valuable if people realize that they can always get a ride with Uber because there's always enough people to fulfill the demand. Now, finally we get to the fail of the week, which unfortunately in this case I have to highlight Gemini. Yes, the Gemini from the Google that I know and love.
In short: Dawson names it as one of the two megacaps whose headline numbers are most exposed to the one-time-gains problem, without a view on the business: "we have the contribution as well from these one-time gains, from the security gains that are skewing all of these GAAP headline earnings numbers… it's one of the reasons why I think as we go into '27, we could be talking about negative year-over-year growth in a quarter like the second quarter, simply because you have a big huge gap of missing one-time gains that are reported in the headline numbers. And the important part of that is that Google, Amazon, they only report on a GAAP basis. They don't report an adjusted number like some companies will."
Some companies report two profit numbers: the official GAAP one, and an "adjusted" one that strips out items management considers one-off. Alphabet reports only GAAP. Normally that is the conservative choice and a point in its favour.
Dawson's twist is that right now it cuts the other way. Big one-time gains on securities — mark-ups on stakes in other companies — are flowing into the headline GAAP earnings and flattering them. Because there is no adjusted number published alongside, the flattering figure is the headline figure.
The consequence is arithmetic, not a judgement about the business: gains that big are unlikely to repeat, so when the comparison quarter comes around in 2027 the year-over-year change could be negative purely because last year's number contained something this year's does not. Worth knowing in advance so a mechanical comp is not mistaken for a deterioration.
24:01reported in the headline numbers. And the important part of that is that Google, Amazon, they only report on a GAAP basis. They don't report an adjusted number like some companies will. And there's a lot of issues with adjusted numbers, but the point is accounting is not something that people care about till the peak of the cycle.
In short: Appears twice in the same Nathan passage, both times as a comparator: as the third of the cloud businesses Meta lacks, and — via Gemini — as one of the frontier models Meta has been "lagging in the performance" of, alongside OpenAI and Anthropic. The argument that follows would cut against any model-based moat, Google's included: "they're all going to be commoditized… you're going to choose the cheapest one from a token perspective."
30:06One of the knocks on Meta is that they basically have been lagging in the performance of their models to OpenAI, Anthropic, even Gemini. And the monetization of all that spend. They don't have that big cloud business like Amazon, like Microsoft, and like Google.
In short: Largest position, $200k (+$105k). Gemini 3.8 Flash, Gemini past 1B MAU, and the DOJ ad-tech divestiture rejected ("two for two" in court). But 15% growth × 25× gives only 9.6% — "below my hurdle… not suggesting Google's a super strong buy today. We need to have a dip."
Google is his largest holding and business news is good — its Gemini AI has passed a billion monthly users and a judge refused to force the sale of its ad-tech business. But at today's price his math gives only about 9.6% a year, below his 15% bar, so he holds it without buying more and wants a dip first.
In short: A single appearance, as a datapoint rather than a position. Lebenthal's evidence that hyperscaler CapEx is being monetised: "Alphabet growing its web services at 80%, Microsoft Azure at 40%. There are fundamental earnings based reasons why the MAG 7 is performing." No stance was given on the stock this hour.
Alphabet is named once, as a number rather than a position: its cloud business growing 80%, alongside Microsoft's Azure at 40%, in Lebenthal's list of reasons the AI build-out is earning a return.
The relevance is to the sector argument, not to the stock. If the companies renting out AI computing power are growing at those rates, the spending that created the capacity is being absorbed by real demand — which is the difference between a build-out and a bubble.
In short: "I think the best AI company out there when you look at the whole ecosystem, which is Google" — the licensor of Gemini to Apple — and "this Gemini Flash when it comes out from Google… is going to be pretty darn good in terms of low-cost token production," so the US won't fall far behind China's cheap models. Google Cloud is in the 35% → 43% accelerating-cloud group. The caveat: "Google hitting the debt markets and the equity markets. They went cash flow negative for the first time since they went public."
Niles calls Google "the best AI company out there when you look at the whole ecosystem" — it has the models (Gemini), the cloud to run them, and the products to put them in. Apple choosing to rent Gemini rather than build its own is the proof in this conversation.
He also sees Google as America's answer to cheap Chinese AI. China's labs, cut off from the best chips, got very good at making AI cheaply. Google's upcoming "Gemini Flash" — a smaller, cheaper model — should be "pretty darn good" on cost, so the US doesn't lose the low end of the market.
The worry is money, not technology. Google now spends so much on AI data centers that its free cash flow — cash left after paying for everything — went negative for the first time since it went public, and it has been raising money through debt and stock markets. Funding a build-out with borrowed money is what makes a bubble dangerous when it ends.
31:44So for right now it has worked out very well for Apple because they also chose to partner with I think the best AI company out there when you look at the whole ecosystem which is Google. And so instead of spending hundreds of billions of dollars, the rumor is that they're spending like a couple of billion dollars to license Gemini for the iPhone from Google, working with Google to get these out there.
In short: Google Cloud is one of the "big three public cloud vendors" that all "saw accelerating revenue growth from the March to the June quarter… from increasing 35%… to up 43%. But more importantly, the operating margins also expanded about 2 percentage points." Structurally he still has "Anthropic… on the enterprise side and then Google on the consumer side… going to squeeze OpenAI between them," and Google is one of the internet-bust survivors he uses as the good outcome ("you don't know if you're holding the next Google or… the next Yahoo"). The caution attached to it is cash flow, not growth: "your cash flows, which had been hugely positive, like for Google, are now negative for the first time since they went public."
Google Cloud is one of the three big rental-computing businesses Niles uses to test whether AI is really making money. His test has two parts: is growth speeding up, and is the profit margin rising at the same time? For the three big clouds together, growth went from 35% to 43% between the March and June quarters and operating margins widened by about two percentage points. Both halves passed.
Google also sits on the winning side of his structural argument about who ends up with the profits. He expects AI to settle into a two-winner market split by customer type, with Anthropic taking businesses and Google taking consumers — squeezing OpenAI in between. His reason is simple and behavioural: people have been trained for twenty years to get answers from Google for nothing, so consumers are extremely hard to charge, and the company that already owns free consumer distribution is best placed to monetise them some other way.
The caveat he raises is about cash, not customers. Alphabet's free cash flow — the money left over after running the business and building the data centres — has gone negative "for the first time since they went public." That does not change the demand picture, but it does mean the build is now being funded partly with borrowing, which is exactly why he watches the debt market as closely as the earnings.
He also uses Google as the shorthand for the good outcome in his survivorship warning: you don't know in advance whether you own "the next Google or… the next Yahoo."
5:21All three of them saw accelerating revenue growth from the March to the June quarter where revenues went from increasing 35% in March year-over-year for their cloud businesses to up 43%. But more importantly, the operating margins also expanded about 2 percentage points from March to June. And so not only were revenues accelerating, but the profitability of those revenues was also going up.
In short: The exhibit for the whole debt-financed-AI argument rather than an equity view: "when Google comes to market with a $40 billion bond issue, that puts a lot of pressure on the market," and it is the concrete case of the thing he has never seen before — "in my lifetime, this is the first time I've seen tech being financed with debt… usually you finance tech with equity, right? Because the asset has a very short lifespan." The cost is the punchline: "I don't know what the spread of Google paper is over treasuries… but they're essentially paying a 6% coupon on this debt. It's a lot."
Alphabet is Google's parent — search, YouTube, Android, and one of the handful of companies spending enormous sums building AI data centres.
In this conversation Alphabet is not a stock idea; it is the evidence for Dillian's bubble diagnosis. Its $40 billion bond sale is his example of something he says he has never seen before in his career: technology being financed with borrowed money rather than shares. His objection is a matching problem — an AI data centre is obsolete in a few years, so funding it with ten- or thirty-year debt commits you to payments long after the asset has stopped earning. Equity has no such schedule; debt does.
The price of that debt is the second half. Google's bonds yield only modestly more than Treasuries, but because Treasury yields are so high, "they're essentially paying a 6% coupon on this debt" against roughly 2% in 2021. And because these issues are enormous, they compete with the government for the same pool of savings — the crowding-out effect Volcker described in the late 1970s, now running in reverse with corporates doing the crowding.
5:17— Yeah. One point I do want to bring up is that a lot of this AI issuance is weighing on the bond market. So when Google comes to market with a $40 billion bond issue, that puts a lot of pressure on the market. So, you were talking about Volcker basically talking about the crowding out effect. Yes.
In short: Greg Abel: Buffett initiated the position "probably close to 15 months ago or a little bit more" and Berkshire kept buying; then in late May Abel took a negotiated block in Alphabet's equity offering — "recommended that we consider 10 billion and more… I'd recommended 6.5% discount." The reason given: "we all are seeing and feeling the impact of AI… and then we saw Google as a significant player… now have a significant investment in it," with "a lot more to Google than what I just said."
Alphabet is Google's parent — search, YouTube, the Android phone software, the Google Cloud computing business, and the Gemini AI models. Berkshire Hathaway, the company Warren Buffett built and Greg Abel now runs, has been buying it for well over a year, and this interview is the first detailed public account of how the biggest piece of that stake was bought.
The mechanics are worth understanding because they are unusual. When a large public company wants to raise cash quickly, it can sell a big new slab of shares to one buyer rather than dribbling them into the open market — and because that buyer is doing the company a favour by absorbing the whole lot at once, the shares are sold at a discount to the market price. That discount is the buyer's compensation for the size and the speed. In late May, Alphabet called Berkshire on a Sunday morning to ask whether it wanted in. Crucially, "no terms or amount were set" — the price and the size were still open. Berkshire named both: at least $10 billion, at a 6.5% discount to the market. Alphabet agreed. So Berkshire started the position roughly six and a half percent below where anyone else could have bought it that day, purely for being big enough to write the cheque in one go.
The reasoning behind wanting more Alphabet in the first place is stated only at a high level — Berkshire deliberately does not discuss the specifics behind its equity positions. What Abel does reveal is where the conviction came from: Berkshire owns roughly a hundred operating businesses, from a railroad to insurers to a utility, and it can see directly what AI is doing inside them — "a lot of visibility from within our companies as to how we're using AI, what type of benefits it's delivering." That is genuine private information about demand, and it is what made a technology company interesting to a firm that historically avoided them. From there the judgement is simple and, he admits, incomplete: "we saw Google as a significant player" in AI, and "there's a lot more to Google than what I just said."
The stance is positive, but it is a positive from a buyer with a structural advantage no ordinary investor has — a discounted entry that came with the size of the cheque. Someone paying the market price today is not getting the same trade.
3:31— Why do you like Alphabet? — I think from just from a real high level, obviously, we don't discuss the underlying specifics of any of the concepts in or around any of our equity investments. But the one thing that is unique with Alphabet, and I guess we do see this across our other businesses, but number one, obviously we all are seeing and feeling the impact of AI.
In short: Steve Weiss's final trade, on a purely cyclical view of AI leadership: "Alphabet — these things cycle in terms of who's got the lead in the AI technology. This is really too cheap not to own right here." The setup: the Wall Street Journal reports a Gemini 3.8 Flash release this week, and a federal judge ruled today that Google does not have to sell its advertising-technology business (most of the DOJ's proposed remedies rejected). Lebenthal, also an owner, supplies the bear half honestly: "they need a little resurgence in their AI model. About nine months ago, they were on the top of the heap. Since that time, Anthropic has clearly taken over in terms of the most popular, most used large language model. OpenAI is not being left behind either… the stock has fallen off from roughly 400 to around 330 today. They need this model to do well in the marketplace." Alphabet is also named twice on the other side of the ledger — as a source of new cyber tools pressuring Palo Alto, and as Broadcom's custom-silicon partner.
Alphabet appears on both sides of the ledger in this episode, which is unusual and worth reading carefully.
The bear half comes from an owner: nine months ago Google had the leading AI model, and since then Anthropic has taken the lead in usage while OpenAI has not fallen behind either. The stock has gone from roughly $400 to $330. Lebenthal's point is that in this market the model ranking is not a vanity metric — it determines who wins enterprise contracts — so the Gemini 3.8 Flash release this week actually matters.
The bull half is Weiss's, and it rests on the same volatility: leadership in AI "cycles," so a company that has lost the crown once can take it back, and at this price you are not paying for the crown. He also gets a clean regulatory win the same day — a judge ruled Google does not have to sell its advertising-technology business, rejecting most of the government's proposed remedies, which removes a large and long-running tail risk from the valuation.
In short: Held, not a buy at ~$342 — "just at a normal valuation… nothing special." Would add "probably around 260… if it goes down [to] 250." Agrees the moat isn't having the single best model but the full stack: good-enough Gemini + cloud + TPUs (less Nvidia reliance) + YouTube/Gmail/Maps + Android devices, which "gives them a million ways to make money."
Google is one of Carlson's largest holdings, but at about $342 a share he thinks it is fairly priced rather than cheap, so he isn't adding; he would consider buying more around $250–260. He agrees with a member that Google's advantage in AI is not having the single smartest model — it just needs a very good one — but owning the whole chain: its own chips, its cloud, apps billions of people already use (YouTube, Gmail, Maps) and the Android phones they use them on. That gives it many ways to make money from AI.
In short: Search still growing double digits and the stock "reasonably cheap" (he owns some megacaps) — but it just printed its first negative free-cash-flow quarter and the sell-side's snap-back to positive FCF in 2028 presupposes ROIC visibility nobody has. Sergey and Larry coming back to work on AI is his evidence the spend is existential, not financial.
Google's core search business is still growing double digits and Keller thinks the stock is reasonably cheap — but it just reported its first quarter ever where the business consumed more cash than it produced, because of the sheer scale of AI spending. Wall Street's forecasts show that reversing and free cash flow turning strongly positive again by 2028, which quietly assumes everyone knows what the return on this spending will be. Nobody does.
His read is that the spending continues regardless. Google's founders came back to work on this personally; the leadership treats it as existential and generational, not as a return-on-capital exercise. That's good for the technology and ambiguous for shareholders: the money keeps going out the door whether or not the payback maths works, and the era when simply owning the biggest tech companies produced extraordinary returns is probably over.
27:28— Now what do you think about — obviously we saw Google report its first negative free cash flow quarter. Is that coming for the rest of the hyperscalers and how do you think markets are going to think about the Mag 7 and the hyperscaler names that were the market leaders — they are definitively not the market leaders right now. Is that trade of just buy the big megacap tech companies, do you think that's over? Not to say that they won't keep up with the market but
In short: The second half of the earnings-optics problem, and a bull case running alongside it. The optics: headline GAAP EPS growth of 118.5% is inflated by "one-time gains on private investments for several large tech companies… given how big Anthropic's mark-to-market was for Amazon and Google's P&L"; strip the pair and the beat rate falls 29% → 10%. The bull case, from the deck: Wolfe Research named Alphabet a top pick for 2027 with a $460 target, modelling $105B+ of TPU revenue next year at ~30% margins (~$32B of TPU operating income), $300B cumulative TPU revenue through FY28, power capacity to 21 GW next year and 28 GW in 2028, and GCP eventually at up to 40% of Alphabet's operating income. Also hired Barret Zoph, ex-CTO of Thinking Machines Lab, as VP of Research on Gemini — and sits inside the $3.1T off-balance-sheet tally.
Alphabet is the other half of the earnings-optics problem. Like Amazon it holds a large stake in Anthropic, and the increase in that stake's paper value flows straight through to reported profit. Headline earnings growth for the S&P 500 reads 118.5%; on an adjusted basis that strips out these one-off revaluations it is 31%. Singh's instruction is to use the second number and treat the first "with a grain of salt."
Running alongside that is a genuine bull case from Wolfe Research, which named Alphabet its top pick for 2027 with a $460 target. The argument is about the chips Google designs itself. Wolfe models more than $105 billion of revenue next year from selling access to those processors at roughly 30% margins — about $32 billion of operating profit — and $300 billion cumulatively through 2028, with Google's electrical capacity scaling to 21 gigawatts next year. On those numbers the cloud division would eventually generate up to 40% of the company's operating profit.
So Alphabet is simultaneously the reason to distrust reported index earnings and one of the few companies with a credible path to justifying its AI spending with actual revenue. Singh does not resolve the tension; he notes both.
Full passage: premium transcript (PDF).
In short: Named only inside Faber's question — does one model winner emerge "as with Ask Jeeves and all the other search engines"? Kedrosky's answer is that no such separation is coming: the models have converged, so the search-era winner-take-all analogy doesn't apply this time.
33:53They're going to come out of nowhere. They're all going to be coming out of nowhere because the process of producing chips itself is being transformed by AI, which in a perverse sort of way will help prick the bubble itself. — Is it a scenario where there's five, 10, 20, 50 of them that are all good, but it's hard to distinguish between one being slightly better than the other? Or is it a scenario where maybe Google bubbles up above, as with Ask Jeeves and all the other search engines, and there is kind of a
In short: Peak-date evidence, not a stance: charted with NVDA as having peaked "back in May." Named only as a coordinate in the leadership scatter — no update to his standing Google view.
In short: Referenced as part of the hyperscaler class, not named individually here. Read alongside the Aug 15 disclosure that he sold his position to cut AI exposure — the "still long several other tech stocks" line in this interview does not identify which.
4:45Which of those companies, we don't know who they are, but we could all pretty much guess a bunch of them. — Mhm. — The way I see the entire AI story is that Nvidia sells chips to hyperscalers. 70% of hyperscaler AI revenue is from Anthropic and OpenAI. And that equates to about 25 to 35% of their cloud revenue.
In short: Held; cited with Amazon as more durable AI beneficiaries than Nvidia — "more infrastructure… more customers."
In short: Buy target $250 (~$100 below the $347 price; his largest position, +$113k). A hold, not a buy, at today's price: "at 347, it's trading at a 26 forward PE ratio and a 1.2% free cash flow yield. I don't believe that Google's cheap today. I'm not buying it right now. I continue to hold it mostly because it's such a good company." The target is the worked example of his method — moderate growth to 12% EPS (because "they're building out mass amounts of data centers… which will transform into amortized expenses") and haircut the exit multiple to 22, and $250 still compounds at 14.1% to roughly $484 in five years.
Google is his largest position and, right now, a hold rather than a buy — a distinction he is unusually blunt about. At $347 the stock trades at 26 times next year's earnings and throws off only a 1.2% free-cash-flow yield: "I don't believe that Google's cheap today. I'm not buying it right now. I continue to hold it mostly because it's such a good company."
Google is also where he shows his whole method. He takes the growth rate Google is actually achieving and cuts it — to 12% — on the reasoning that the enormous data-centre build will eventually be charged against earnings as those assets are written down over time. Then he cuts the multiple investors might pay, from today's 26 down to 22, on the assumption that a slower grower earns a lower rating. Both haircuts are deliberate: he wants a target that works even if the story gets worse.
With those inputs he solves backwards for the price that still delivers a good return, and it lands at $250 — roughly $100 below the market. Buy there and the same pessimistic assumptions still produce a 14.1% annual return, taking the stock to about $484 in five years. That's the whole discipline in one worked example: the target is not a prediction of where the stock goes, it's the entry price at which a deliberately gloomy forecast still pays you well.
4:17I continue to hold it mostly because it's such a good company and I believe the future is bright for it, but overall it's not a company that I think is a screaming buy. And the buy target that I've set for Google is $250, so around $100 less than its current price. Let's go ahead and bring up the DCF calculator and we can do some quick math here.
In short: The lead exhibit in the big-tech bond paradox: capex used to be funded by free cash flow, "but that dried up. So now it's by issuing bonds. So Google issued a 100-year bond" (at roughly a 6% yield) — paper that "if you are a credit investor and don't think that Meta and Google are going out of business," you buy instead of Treasuries, "causing yields to go higher." Separately, supply-chain de-risking: Google "plans to move production of all Pixel smartphones, watches and earbuds out of China starting in 2027, mainly to Vietnam and India."
Alphabet appears here as the lead example of what Singh calls the big-tech bond paradox. It has issued a 100-year bond — borrowing money repayable a century from now — at roughly 6%. The proceeds fund data centres.
Why this matters beyond Alphabet: pension funds and insurers have to put money somewhere safe for the long term, and have traditionally bought US government bonds. Alphabet's paper pays two to three percentage points more, and few credit investors believe Alphabet is going bankrupt. So money leaves government bonds for corporate bonds, which pushes government borrowing costs up. The paradox is that the AI investment supposed to grow the economy out of its debt problem is being financed by taking money away from the government's own debt.
Separately, Alphabet is moving all Pixel phone, watch and earbud manufacturing out of China to Vietnam and India from 2027 — a supply-chain hedge against exactly the kind of tariff escalation the rest of the call describes.
Full passage: premium transcript (PDF).
In short: Another walled garden Zeta integrates with and activates on (Google, YouTube); cited alongside Meta, Amazon, Netflix and Spotify as names that would share any privacy-regulation hit — "it doesn't just affect Zeta in this box."
In short: UPGRADED HOLD → BUY — "Google's parent company focused on internet search, AI, and digital advertising." The rating is unusual because two of the three models disagree with it: fwd PE 29.9 against a 22.4 five-year average, i.e. 33.5% overvalued, and the RDCF requires 22.6% growth against 15.0% expected (−7.6pp). Only the earnings-growth model supports it, at a 13.66% expected return. YTD +8.3%, ten-year CAGR 24.2%.
Alphabet is upgraded from hold to buy this month. It is worth being clear about what that upgrade rests on, because two of the three valuation methods the letter publishes argue against it: the shares trade at about 30 times forward earnings against a five-year average of 22 — expensive by their own history — and the reverse-DCF test says the price already assumes 22.6% annual growth while only 15% is expected.
Only the third method supports it, projecting a 13.7% annual return from growth plus dividend plus a modest change in multiple. No written reasoning is given beyond a one-line description of the business. So this is a judgement call presented alongside a spreadsheet that disagrees with it, and a reader should treat it that way rather than assuming the numbers backed it up.
In short: Raised by the host, not by Gromen: Google's reported "800 billion dollars in forward purchase commitments" as evidence that the Q4 numbers will confirm what the off-balance-sheet agreements already show. Gromen concedes the near-term point — "I agree with you that it's going higher, the bubble hasn't popped yet" — while insisting every prior capex boom ended in a bust.
1:10:17But I think that Google, as FT, Wall Street Journal reported, they have 800 billion dollars in forward purchase commitments. This stuff is going to happen. So I think my confidence that in Q4 when these semiconductor companies or the power electrical companies report, it's going to reflect the reality that we already see in these off-balance sheet agreements.
In short: Simpson bought more — but explicitly not on the Marvell headline: "completely unrelated to this, Scott. But I do like the diversification they're doing within their TPU ecosystem — they've got Broadcom, they've got NVIDIA, now they've got Marvell, and it's just shoring up what they need." On the long-standing criticism: "if you would criticise Alphabet for anything, it's the massive spend. But what we're seeing now is a bit more of a disciplined and targeted spend." The position is being built deliberately and as a source of funds is IBM: "we've been building this position out here regardless of this move… I look at this as a really long-term hold. If you think about it, we basically went from IBM into Alphabet — it's been taking a few months for us to do it. We still have a 1% position in IBM, but basically that's what we're doing." Over two to three years he calls Alphabet "a much better play, more diversified business."
Kevin Simpson bought more Alphabet, and was careful to say it had nothing to do with today's Marvell headline. What he likes is that Alphabet now has three separate suppliers for the chips that run its AI — Broadcom, Nvidia and now Marvell. Depending on one supplier is a hostage position; depending on three is a negotiating position.
The other change he cites is discipline. The standing criticism of Alphabet has been the sheer size of its AI spending; his read is that the spending has become more targeted, which is what an investor wants to see before a capital-heavy phase turns into returns.
Most usefully, he names the source of the money: he is moving out of IBM and into Alphabet, a switch that has taken months and is nearly finished (IBM is down to a 1% position). Over a two-to-three-year horizon he calls Alphabet "a much better play, more diversified business" — and this is a long-term hold, not a trade.
In short: The most-contested mega-cap of the quarter, and net strongly accumulated. Alphabet is the third-most widely held name — "Alphabet follows with eight" of the 20 funds' top-five holdings. Berkshire "boosted GOOG/L by 83%, making Alphabet its third-largest stock holding behind Apple and American Express" — roughly the centrepiece of its first net-buying quarter in 14. Druckenmiller "reopened Alphabet with 336,000 shares" after completely exiting in Q1. Against that, Ackman "finished exiting GOOGL, completing the rotation that started last quarter."
Alphabet is Google — search, YouTube, Android, the Gemini models and Google Cloud. It spent much of the last two years as the mega-cap investors argued about most, on the theory that AI chatbots would erode search.
Q2's filings show that argument resolving in one direction. It is the third-most widely held name (a top-five position at eight of the 20 funds); Berkshire increased its stake by 83%, making Alphabet its third-largest holding behind only Apple and American Express; and Druckenmiller reopened a 336,000-share position after having exited entirely one quarter earlier.
The dissenting vote is worth keeping in view: Ackman finished selling out of Alphabet, completing a rotation he began the previous quarter. Two respected managers took opposite sides of the same name in the same three months — which is the practical argument against treating any one 13F as a signal.
In short: "I still like Google" — Q2 revenue +24%, operating income $41B, Search +17% on AI Overviews, YouTube ads $11.1B (+13%), subscriptions +15%; the stock didn't rally because $45B of quarterly capex took non-GAAP cash flow to −$5.85B and EPS was skewed by the Anthropic stake. The caveat: it is no longer the capital-light business it was.
Alphabet reported a strong quarter — revenue up 24%, Search up 17% helped by AI Overviews, YouTube ads and subscriptions both growing — and the stock still didn't rally. The reason is $45 billion of capital spending in a single quarter, which pushed the company's quarterly cash flow to negative $5.85 billion.
Singh's takeaway isn't that Alphabet is bad — "I still like Google" — but that it is no longer the kind of business it was. It used to print cash almost effortlessly; now it consumes it, so the old valuation habits no longer apply. Its stake in Anthropic also flatters its reported earnings without any cash changing hands.
17:34So, that was quite positive. The reason why the stock didn't rally a ton, despite even YouTube ads being up 11.1 billion up 13% and subscriptions up 15% was that they did 45 billion capex for the quarter, bringing quarterly non-GAAP cash flow to -5.85 billion. And so, because of that, what we have seen is that the market is penalizing tech companies for effectively overspending on capex.
In short: The other side of Hohn's rotation — Google added 12% to roughly an 8% combined position, and the exact reverse of Hohn's 2018 trade (he sold Google then on ChatGPT-disruption fear and bought Microsoft). Carlson's read: "Google's moat wasn't disrupted… many investors have come to the same conclusion that Google is still wide moat." He owns both sides of the trade: "Whether or not you're flipping the coin between Microsoft and Google, they're both great. I own both of the companies. I'm bullish on both of them."
Google is where Hohn's Microsoft money went — a 12% add taking the combined stake to roughly 8% of a $52.7B book. What makes it instructive is that it exactly reverses his 2018 decision, when he sold Google because he feared ChatGPT would break search, and bought Microsoft instead.
He was wrong then, and he changed his mind on evidence rather than narrative: the disruption "did not manifest in the numbers." Carlson notes the market has now reached the same conclusion, which is why Google has already re-rated to a much higher valuation than when Hohn was buying.
For Carlson himself there's no new call here — he owns Google and Microsoft both, and says flipping a coin between them is fine because both are great businesses. The takeaway he draws is about process: a manager willing to publicly reverse a headline thesis when the data contradicts it is doing the job correctly.
13:42He believes that a lot of Azure is reliant on Microsoft Office. So, it's kind of like a domino effect. If Microsoft Office falls, Azure will suffer as a result. As a result, he's out of Microsoft and he has Google at a combined position of around 8%. He added a 12% to his position in the last quarter. So, he continues to build conviction on Google in this most recent quarter.
In short: The name that got Strategas fired. Trennert's "anodyne sentence" — "it's going to be hard to continue to grow AI investments without further dilution of shareholders" — was written "mainly cuz you saw Google issue 85 billion dollars in equity," and he calls the raise "shocking given the fact that the last time they actually raised equity capital was in the 2000s." Against that, Verrone's tape: "Google's basically back at the highs here." Net: dilution is the structural worry, momentum is the near-term fact — and Trennert explicitly "wouldn't short it."
Alphabet is the name at the centre of the episode's most human story. Strategas published a sentence — "it's going to be hard to continue to grow AI investments without further dilution of shareholders" — and a client fired Trennert personally over it.
"Dilution" simply means issuing new shares: each existing shareholder ends up owning a smaller slice of the same company. Alphabet raised $85bn of equity, and the reason that is startling is history — a company that has thrown off cash since the early 2000s has not needed to ask shareholders for money since then. The AI build-out has turned a cash machine into a business that taps the capital markets.
Against that, Verrone notes the stock is "basically back at the highs." The net stance here is genuinely mixed, and Trennert says so explicitly: the dilution worry is real, he expects the spending to continue, and "I wouldn't short it." He is market-weight technology — neither over- nor under-owning it relative to the index.
21:50I don't think. The fact that Google raised $85 billion in equity is shocking given the fact that the last time they actually raised equity capital was in the 2000s. Right. Early 2000s. Yeah. So, to me, I wouldn't short it. We're market-weight technology and I do think the demand is exceeding the supply for artificial intelligence.
In short: Two contradictory readings in one call. The distribution win: "Gemini has become Google's fastest growing product ever, crossing 1 billion monthly users" — from 400M in May 2025 — though "because they're basically forcing people to use it." The accounting warning: Alphabet's Q2 GAAP EPS of $9.11 against a $2.88 estimate "included a gain of 98 billion in other income primarily due to unrealized gains on equity securities like Anthropic," and stripping Alphabet and Amazon takes the index's earnings-surprise percentage from 29.2% down to 11% and blended EPS growth from 50.4% to 32%. Deck page 40.
Full passage: premium transcript (PDF).
In short: The best discloser of the group — "admirable… you can search for it and find it fairly easily," and the only one he could read without a long dig. The number is the largest: $800bn of purchase commitments (chips, memory, equipment, cooling, electricity), almost half the hyperscaler total, of which ~$200bn is flagged short-term — which, if it means the next 12 months, implies published capex expectations are too low. His guess on why: "maybe their accountants got a little bit worried."
A "purchase commitment" is a contractual promise to buy something in future — here, chips, memory, cooling equipment and, increasingly, guaranteed electricity. It isn't debt, but it is money you have already agreed to spend, and Wigglesworth found $800 billion of it at Alphabet: nearly half of the entire hyperscaler total he could tally.
Alphabet comes out of the exercise well on process and badly on scale. It is the one company whose filings he could actually read — "admirable… you can search for it and find it fairly easily" — while everyone else buries the same information in inconsistent language across different sections. His half-joking explanation is that the number got big enough that the accountants decided transparency was the safer course.
The investable observation is the maturity split: roughly $200bn of the $800bn is labelled short-term, which he reads as due within about twelve months. If that is right, the market's published capex estimate for Alphabet is too low — the spending is contracted, it is just not yet in the forecast.
6:33— So that's chips, memory, equipment, cooling, electricity, the whole jamboree. — But that's almost half the total. I think maybe their accountants got a little bit worried and thought we need to be quite transparent about this. — And do you have a rough sense of over how many years the majority of that is scheduled for? Because if that's scheduled for the next three years then the Google capex number for expectations is too low.
In short: Referenced — roughly $10B of Alphabet was the named component of Berkshire's $23B of Q2 equity purchases, its first quarter as a net buyer of stocks in more than three years. A capital-allocation datapoint inside the Berkshire section, not a fresh view on Alphabet.
In short: Referenced twice, both structurally: as Waymo's balance sheet ("they have Google's backing, so they have a lot of capital to be able to invest in this"), and as an aggregator archetype alongside YouTube and the Google Play Store — "when other things are commoditized and there's lots of supply, the aggregators usually win."
In short: His example of the capex question investors should actually be focused on: "you see Google trying to raise 25, 30 billion dollars and they're already spending over 200 billion on CapEx. These are extraordinary numbers. This isn't going away anytime soon." His verdict is not bearish — "I don't sense that it's a bubble. These companies are making really an extraordinary amount of money" — but it is where he says the risk lives over the next two years.
Alphabet appears as Newton's illustration of the one thing he thinks investors should be watching: the sheer scale of AI capital spending. Google is "already spending over 200 billion on CapEx" and is out raising another "25, 30 billion dollars" in debt. "These are extraordinary numbers."
Capital expenditure is money spent on physical assets — here, data centres and chips — which is paid out now against revenue hoped for later. When a company that generates enormous cash still needs to borrow tens of billions on top, the spending has outrun even its own cash flow, and the returns become the market's central question.
His conclusion, though, is not bearish: "I don't sense that it's a bubble. These companies are making really an extraordinary amount of money." He compares the moment to the late 1990s — a real, profitable boom that ran for years before it broke — and puts the window he would actually worry about at 2028 to 2030, not now.
36:15I think that most of the concerns that investors have should be more focused on the level of CapEx and really what's going on to fuel the AI trade that really is going to continue to dominate the narrative likely over the next 2 years. I think when you have companies that are trying like hell to raise money and you see Google trying to raise 25, 30 billion dollars and they're already spending over 200 billion on CapEx.
In short: Sold. "I sold my Google. I wanted to reduce my exposure to AI. What scares me is that it's all one trade. So it better succeed." Not a company complaint — elsewhere in the same clips Google is one of the hyperscalers with "multiple revenue streams from established businesses." The sale is an exposure decision, not a verdict on the business.
Google is a hyperscaler: alongside search and ads, it rents out enormous data centres that other companies' AI models run inside. Eisman is not saying the business is broken — in the same set of clips he lists Google among the companies with "multiple revenue streams from established businesses which are very unlikely to simply disappear."
He sold it anyway, and the reason is about his portfolio rather than the company: "I sold my Google. I wanted to reduce my exposure to AI." His worry is that almost everything in a normal portfolio now rises and falls with the same AI story, so owning a great AI-linked business does nothing to spread risk. Selling a name he still respects is the price of cutting a concentration he thinks is invisible to most investors.
0:26So much so he's getting out of his AI plays. — I sold my Google. I wanted to reduce my exposure to AI. What scares me is that it's all one trade. So it better succeed. — But what is he actually saying that the rest of the market is only just waking up to? Well, there are three dominos that Steve puts forward.
In short: Named with Microsoft and Amazon as one of the three "hyperscaler winners." Carries the same embedded risk: 25-35% of total cloud revenue traces back to Anthropic and OpenAI, so "should Anthropic and OpenAI fail, the margins of the hyperscalers will compress and business will return to slower growth."
Google Cloud is the second of his three hyperscaler winners, and it carries exactly the same shape of risk: a big, growing, genuinely defensible infrastructure business whose fastest-growing customers are the two AI labs.
Notably this is a softer stance than a few weeks ago, when he said he had sold his long-held Google position to reduce AI exposure. Here the company is on the right side of the divide he's drawing — the layer with a moat rather than the layer without one.
12:49This bare case of entropic and open AI could occur but it might occur a year from now and for markets that is an eternity. Winners and losers are beginning to emerge amongst the big companies. Microsoft, Google and Amazon seem to be the hyperscaler winners. Oracle is questionable with its debt barely rated above junk. Meta is having problems.
In short: Disclosed as a continuing holding in the "I'm not copying his trades" list — "I had Google before he bought Google." Named as one of his own positions that Ackman later joined; the thesis isn't re-argued in this episode.
1:35I've already been in a lot of these companies. I bought Meta before it was revealed that Bill Ackman bought Meta. I had Google before he bought Google. I had Microsoft before he had Microsoft in his portfolio. I had Amazon before he bought Amazon. I had Netflix twice before he bought Netflix. And the same goes for now MasterCard and S&P Global.
In short: #2. "They dominate digital advertising, cloud computing, and consumer tech through products like Search, YouTube, Android, and Maps." Durability: "People will always need a reliable way to search, index, and access global knowledge; they possess an irreplaceable data advantage for AI; Google services are woven into daily human habit." Third in the June 2026 market-cap table at $4.194trn. Bought by Kris Heyndrikx at the value stage of the same de-rating two days earlier in this archive's growth interview — the two arguments arrive at the same name from opposite styles.
Alphabet owns Google search, YouTube, Android, Maps and a large cloud-computing business, and makes most of its money selling advertising against what people are already looking for. That is the most valuable kind of advertising there is, because intent is already present.
Three reasons are given for the next twenty years: people will always need a way to find things; the accumulated record of what billions of people search for is a training advantage for AI that cannot be bought; and the products are woven into daily habit. The obvious counter — that AI assistants replace the search box entirely — is not addressed here, though two days earlier in this archive a growth investor described buying the shares precisely because the market had over-believed that story.
In short: Its $85B upsized equity raise — the largest in US corporate history — to fund AI infrastructure is his supply-overwhelms-demand exhibit: "the ducks are quacking so they're going to feed them." Asked whether the $85B is productive: "the results in the short term have shown you that it's a bad use of capital." Record equity issuance eventually overwhelms demand.
Alphabet just sold $85 billion of new stock — upsized from $80 billion, the largest equity raise in US corporate history — to fund AI data centers. Selling new shares means every existing shareholder owns a slightly smaller piece of the company, so it only makes sense if the money earns a good return. Asked what he'd look for in the filings to judge that, Hayes cut it short: "the results in the short term have shown you that it's a bad use of capital."
His broader worry is arithmetic. The big tech companies used to be the largest buyers of their own shares, which shrank the supply and pushed prices up. Now they've burned through their spare cash and are doing the opposite — issuing record amounts of stock and debt (Intel diluted holders about 20%). "The ducks are quacking so they're going to feed them." At some point that flood of new paper overwhelms the money available to buy it, and prices fall for reasons that have nothing to do with how good the businesses are.
40:41Now, what the market is betting on and the market is a discounting mechanism is that Mñana is going to be a brighter day. And I'm kind of in that camp. I just think that the runway to that brighter day is a bit longer than the market thinks it's going to be right now. and in that kind. So if you look at the semiconductor trade in the last big generational tech runup from 1995 to 2000 during that period there were three, count it, three 50% drawdowns in the semiconductor, the socks index, from 1995 through 2000. So all of the things that
In short: Hasn't traded well relative to the others lately; reiterated at Goldman Sachs today with a 435 target, which flags the brain-drain reorg, the commentary around it, and capital raising in both debt and equity as overhangs. Belsky: "we think they're in the penalty box here a little bit near term because of the reorg, because of where they're spending money. But we still believe that with respect to how they're going to monetize AI, it's proven — they've proven that they can monetize what they run inside the company just like Microsoft. So I think this pullback is a great opportunity for longer-term investors to continue to buy the stock. We think it's one of the five stocks in the US you should absolutely, positively own." And the diversification point: "it's not just one business. They lead in cloud, they lead AI models, YouTube TV, Waymo — it is a diversified business."
Alphabet has lagged its peers, and Goldman's note lists why: the DeepMind reorganisation and researcher departures, the commentary around them, and the fact that Alphabet is now raising capital in both debt and equity markets to fund its AI spending.
Brian Belsky accepts the near-term problem and rejects the long-term one: "they're in the penalty box here a little bit near term because of the reorg, because of where they're spending money." But the key question — whether a company can actually make money from AI — he considers already answered: Alphabet, like Microsoft, has proven it can monetise AI inside its own products. So "this pullback is a great opportunity for longer-term investors… we think it's one of the five stocks in the US you should absolutely, positively own."
His supporting argument is diversification: this isn't a bet on one product. "They lead in cloud, they lead AI models, YouTube TV, Waymo."
In short: Terranova: "the momentum is actually waning in Alphabet since the announcement last week about losing the chief science officer" — the AI brain drain. In a huge week for tech and the Nasdaq, "Alphabet is the only one that's down." His process response: recognize you're holding a name losing momentum, ask whether to hedge — "the answer as it relates to Alphabet is no." Firestone: it had "a really big run from March to the recent high and now it's come back… people will get concerned, those sell some shares, then they'll start to buy" again. Also the destination of Berkshire's $10B secondary purchase.
Alphabet was the only big technology stock down during a huge week for the Nasdaq, and Joe Terranova pins it on the announcement that Alphabet lost its chief science officer — the "AI brain drain," where rival labs poach the researchers whose names justify a company's AI premium.
What's instructive is his process rather than his conclusion. He notices a holding is losing momentum, asks himself whether he wants to hedge it (buy protection or offset it elsewhere), and answers no — the fundamentals still support holding. Karen Firestone reads it as an ordinary round trip: a big run from March to the recent high, a pullback, some nervous selling, then buyers returning. Berkshire's newly disclosed $10 billion purchase sits on the other side of the ledger.
In short: Both a graded past call and a live defence. The past call: accumulated through the stuck $150–180 range (biggest buys ~168) into "a massive position," built by "correctly understand[ing] the bear case… and why it doesn't make sense." The live one: he takes the other side of SemiAnalysis' "Gemini is cooked" article, which says DeepMind's talent exits (Demis, Jeff Dean) drop Google's odds of reaching state-of-the-art again "to zero." Carlson: the same shop called Gemini the best model in the world in November — "if Google is able to create a state-of-the-art model not even 1 year ago, then that means there is some chance that they could do it again… These four or five people made that impossible[?]" He does agree with the report's second half: Google is fine even at second tier because it is deliberately putting more money into Google Cloud than into training models — cloud that "could grow above 100%" with operating margins ticking up "every single quarter. Now up to 35%." "I would not rule out Google so hastily."
A respected research firm, SemiAnalysis, published a piece arguing that senior researchers leaving Google DeepMind mean Gemini will never again be the world's best AI model — and stated the odds as literally "zero," not "close to zero." Carlson takes the other side, on a simple consistency point: the same firm called Gemini the best model in the world less than a year ago. If Google could do it then, from behind and as the underdog, the probability of doing it again cannot be zero.
But he agrees with the article's second half, which matters more for the stock: Google is fine even if its models are permanently second-tier, because it is deliberately shifting money away from training models and into Google Cloud — the business that rents computing power to other companies. That business could roughly double in size over the coming year, and it is becoming more profitable as it grows, with operating margins now around 35%. Given a choice between owning the best model and owning a large, insulated, high-return cloud business, he'd take the cloud.
He also uses Google as the case study for his own method: he accumulated it while it was stuck between $150 and $180 and sentiment was poor, with his biggest purchases around $168. The skill wasn't spotting a good company — it was "correctly understand[ing] the bear case… and why it doesn't make sense."
21:39And I find this hard to believe for a couple of reasons. One of them is that SemiAnalysis themselves, just November of last year, stated that Google's Gemini was the state-of-the-art model. It was likely the best one in the world. And now, not even 1 year later, they have declared that Google can never again do the thing that they did just a year ago, not even a year ago.
In short: Added on the Jeff Dean departure, which took the stock down 5% — "a 250 billion dollar market cap loss for one engineer, seemed kind of silly… it's just one data scientist." The stock "was a little bit extended, so we didn't add a lot." Fundamentals support it — GCP revenue accelerated 63% → 82% with margins +3% sequentially, "crushed AWS performance," and Dan Niles has it as a long-term consumer-AI winner with a complete stack. The caveat is accounting, not operations: of Q2's $9.11 EPS only $2.88 was operating, and Alphabet is now raising $25B of debt for data centers.
Google's chief scientist Jeff Dean announced he was leaving to start a new company, and the stock fell 5% — about $250 billion of value. Singh bought into that, on the simple view that no single researcher is worth a quarter of a trillion dollars: "it's just one data scientist."
The business supports it. Google's cloud arm grew 82% year on year, up from 63% the previous quarter, and expanded its profit margin faster than Amazon's cloud did. He kept the add small only because the stock had already run.
The caveat he raises in the same call is about accounting, not operations. Of the $9.11 per share Alphabet reported, only $2.88 came from actually running the business. The rest was paper gains on its stakes in Anthropic and SpaceX being written up in value — money it never received and could easily have to write back down.
Full passage: premium transcript (PDF).
In short: Two-sided. Positive: Talkington says Amazon and Google "confirm the same thing" as Microsoft — real revenue monetization in cloud. Negative: her praise of Microsoft is explicitly at Google's expense ("they did a great job of reading the room, of not doing what Google's doing — saying they're going to be free cash flow, that they're not going to the debt market"), and she extends this week's DeepMind departures into a governance point: "Microsoft and Google both have a bureaucracy issue… that's why you have to continue to watch the departures leaving to go do their own thing. That bureaucracy has hindered Microsoft, it's definitely hindered Google. You can't ignore these departures, which we all kind of ignore — but I don't think we should."
In short: Dean. Grouped with Microsoft as a rational allocator — spending is disciplined because it monetises through the cloud — and as one of "the usual suspects" that will host enterprises' open-weight inference. Same 10× EV/sales caveat as the rest of the cohort; no separate view on the stock.
54:00— We believe capital allocation with Meta is a bit different than Microsoft and Google who are in the cloud business. So that spending is a lot more measured and a lot more rational. And you could say for instance Meta is doing it as a defensive measure against Tik Tok and other incumbents that could build out GPUs and then they hit some type of app where it's magical and AI is being produced on there and it attracts people's attention which directly hurts Meta. So you could say that it's
In short: Two mentions. Baruch's compute-constraint pillar: "Alphabet even said on their earnings call that they're going to purchase more compute while they build out their compute." And Brown's earnings-quality point: Alphabet's (and Amazon's) reported numbers include write-ups of their Anthropic and startup stakes, "not obviously repeatable" — strip them out and growth is still 28.8%. Baruch also notes he bought more Berkshire right after the Buffett/Alphabet-stake news. No individual call.
In short: The named exhibit for the earnings-quality charge, and the sharpest number Haymaker has published on it: "profits at the big hyperscalers like Alphabet/Google (GOOG) have been greatly flattered by gains on share ownership in their largest AI customers and partners. In GOOG's case, approximately 75% of its first-half 2026 profits have come from the mark-up of its stakes in, primarily, Anthropic and SpaceX." That escalates the "almost two-thirds" figure given to Thoughtful Money on Jul-26 and puts a company-specific number on Aug-3's "extremely flattering accounting, bordering on circular financing… legal but… extremely unsustainable." The implication is left for the reader and is unflattering: the reported "E" is largely a mark-to-market on two private valuations — including SpaceX, whose market value Haymaker documented collapsing from ~$2.5T to $1.4T on Aug-3 — so the same line item reverses when the round-trip goes the other way. A caution on earnings quality, not a rating change: no price, target or portfolio action is attached, and Alphabet remains on Haymaker's Trims/Holds list from the Feb-20 "very spendy" trim.
Hay is not arguing that Google's search or cloud business has broken. He is arguing that the profit number everyone is quoting is mostly not from running a business at all.
Here's how it works. Google has invested money into companies it also sells to — Anthropic (the AI lab) and SpaceX are the two named. Those companies are private, so there is no market price for the shares; their value gets reset whenever a new investor buys in at a higher price. When that happens, Google is allowed to write its stake up to the new figure and record the increase as profit, even though nothing was sold and no cash arrived. Hay's figure: roughly 75% of Alphabet's profit in the first half of 2026 came from these write-ups, not from operations. Three months ago he put the same figure at "almost two-thirds," so it is getting worse, not better.
Why that matters to a shareholder: the write-up works in both directions. If the next funding round for one of those companies is at a lower valuation, the same accounting line turns into a loss of the same magnitude. Hay had already shown one of the two doing exactly that — SpaceX's implied value fell from about $2.5 trillion to $1.4 trillion within weeks. So the earnings that make the stock look reasonably priced are, in effect, borrowed from two private valuations that can move violently.
The wider point of the piece is that this problem is not confined to Google. Strip out energy and everything AI-related and, per economist David Rosenberg, the whole S&P 500's profit growth this year would be about 4% instead of the strongest since 1955. Hay's conclusion is a base rate rather than a prediction: every past spike in corporate profits has been followed by a comparable slump, so expect some reversion — and be suspicious of anyone explaining why it can't happen this time, which is what Irving Fisher was doing nine days before the 1929 crash. Treat this as a caution on the quality of Alphabet's reported earnings; no rating, target or sell instruction is attached.
In short: Down 3.5–5% on a DeepMind reorg (Mackenzie Sagalos): Demis Hassabis moves to a chairman role focused on frontier research, AGI implications, AI safety/policy and a bigger role at Isomorphic Labs; Koray Kavukcuoglu — already running much of the lab's model research as chief AI architect — takes day-to-day leadership; and 27-year chief scientist Jeff Dean leaves to start a company Google will invest in and that will partner with Google Cloud (amicable; the role won't be refilled). It follows John Jumper to Anthropic and Noam Shazeer to OpenAI. The desk is unanimous it's an overreaction. Terranova: "cloud revenue growth 82%, revenue growth 24% — that's what's important to me… a knee-jerk reaction… a shifting in intellectual capital. That's normal." Lebenthal: "there is a lot more talent in the overall AI industry than just one or two or three individuals… Gemini has gone through many versions. This is not a one-man show" — and points to the chips, search, YouTube and Waymo. Snipe: cloud growth 82% vs 60-plus last quarter is "significant runway… this to me is a viable opportunity," and "the deputy is taking over."
Alphabet fell as much as 5% because of people, not numbers. Google reshuffled DeepMind — Demis Hassabis moved up to a chairman role focused on long-range research and AI safety, his deputy took over day-to-day — and, more strikingly, chief scientist Jeff Dean left after nearly three decades to start his own company (which Google will invest in). Two other senior researchers had already gone to Anthropic and OpenAI. Investors read it as a brain drain.
The whole desk called that an overreaction, and each for a different reason. Joe Terranova points at the business: cloud revenue growing 82% and total revenue 24% — "that's what's important to me," and a shuffle of intellectual capital after a strong run is normal. Jim Lebenthal argues the AI industry has far more talent than any two or three individuals, Gemini has already been through many versions, and Alphabet is not one product anyway — it makes its own chips and owns search, YouTube and Waymo. Jason Snipe notes cloud growth accelerated from 60-something percent last quarter to 82%, which is "significant runway," and the person taking over sat right next to the strategy: "a viable opportunity."
In short: Same distortion: a big chunk of the latest earnings came from booking gains on its Anthropic stake — "if you believe Anthropic might be overvalued, Google's earnings are overvalued by default," and it's one-time, not recurring. She endorses the whole hyperscaler critique (FCF negative, debt issued at widening spreads). Her one constructive note is on the credit, not the equity: an investor no longer has to buy a Treasury when Google may be the better-quality credit at a higher yield.
Google is in the same position as Amazon: a chunk of its most recent reported earnings came from marking up its investment in Anthropic, not from selling more advertising or cloud services. As Pomboy puts it, if Anthropic is overvalued, then Google's earnings are overvalued by default — and either way, it's a one-time item that won't repeat next year. Strip that out and the "good" price-to-earnings ratio investors are paying up for looks much less good.
Her second point is the financing. The hyperscalers as a group have gone from generating enormous surplus cash to spending more than they earn, so they are issuing bonds in size — and the market is charging them a widening premium for it. Interestingly, she flips this into a compliment about Google's debt, not its stock: an investor today doesn't have to buy Treasuries, because Google may be as good a credit as the US government while paying a higher yield. That choice is precisely what is pushing government bond yields up.
13:09And basically revealed that yes, they are making these profits right now, but a lot of these profits are either one time. So, like in the case of Google a big chunk of its latest earnings came from booking gains in its stake in Anthropic, right? So, if you believe Anthropic might be overvalued, well, then, Google's earnings are then overvalued by default, and that's a one-time benefit.
In short: Named (with Anthropic) as one of the cloud-compute customers already buying SpaceX's AI capacity — the proof that the $15.8B of AI CapEx in a single quarter has real revenue attached, even while the segment loses $1.3B and SpaceX spends "several dollars today for every dollar of AI revenue." A hyperscaler with its own buildout renting third-party AI capacity is itself the tell about how tight capacity is. (Referenced; not a stance call.)
In short: The standout on his acceleration-plus-margin test: "the big winner, quite honestly, is Google… their revenue growth accelerated 19% and the margins expanded by 3%. And by the way, they're growing 82%. That's a huge number. And so you're seeing both growth and profitability at that layer. And I really like that." Its only handicap was timing — Google "reported about a week before situational awareness got solved."
Niles tests cloud businesses on two things at once, and Google Cloud is the only one that cleared both by a wide margin. The first test is whether growth is speeding up rather than merely fast — its revenue growth accelerated by 19 percentage points, and it is now growing 82%. The second is whether the profit margin is rising while that happens — margins expanded 3 points. "You're seeing both growth and profitability at that layer. And I really like that."
Why both matter together: a cloud provider can always buy faster growth by cutting prices or absorbing the cost of the chips, and the margin is where that shows up. Growth that accelerates while the margin widens means customers are paying up for capacity that is genuinely scarce — the supplier has pricing power, not just volume.
It also sits on the right side of his bigger argument. Google runs one of the platforms that will automatically send each customer request to whichever model is cheapest and good enough. The customer never picks a model vendor; the platform does. That makes the platform the toll booth and the model a substitutable input — and it is a business Google owns end to end, chips included.
One caveat he raises is about timing, not the business: Google reported roughly a week before the market overhang he was waiting on cleared, so the print landed into a worse tape than it deserved.
3:48THE BIG WINNER, QUITE HONESTLY, IS GOOGLE. GOOGLE. BAD NEWS WAS THEY REPORTED ABOUT A WEEK BEFORE SITUATIONAL AWARENESS GOT SOLVED, RIGHT? BUT THEIR REVENUE GROWTH ACCELERATED 19% AND THE MARGINS EXPANDED BY 3%. AND SO AND BY THE WAY, THEY'RE GROWING 82%. THAT'S A HUGE NUMBER. AND SO YOU'RE SEEING BOTH GROWTH AND PROFITABILITY AT THAT LAYER.
In short: Investors "surprisingly sold off" an incredible earnings report, but the stock is getting a lift from Amazon's print (+3% on the day) and is "back to $70,000 of gains" in the Passive Income account. At $223,000 it is still his largest position — "Google and Amazon in my portfolio are in a neck-and-neck race… I feel good about having both of these positions be heavily overweighted."
Google put out what Carlson calls an incredible earnings report and investors sold it anyway — the same pattern as Meta. Now that Amazon has shown what the heavy AI spending actually buys, Google is getting a lift too (up 3% on the day) and his position is back to $70,000 of gains.
At $223,000 it remains his single largest holding, in what he calls a neck-and-neck race with Amazon for the top spot. He is comfortable being deliberately overweight both — the concentration is intentional, not accidental.
11:03Google's still over Amazon at $223,000. Amazon again is at 207. So, right now, Google and Amazon in my portfolio are in a neck-and-neck race to see which company can be a bigger position. And I feel good about having both of these positions be heavily overweighted. Now, throughout last week and into this week, we also have some companies that reported earnings that haven't fared quite as well.
In short: In the drag cohort but on the disciplined side of it: capex guidance raised from $180-190B to $195-205B with negative FCF for the quarter, yet "Microsoft and Google have had largely sustaining AI capex buildout while remaining free cash flow positive" and Google Cloud is a $99B run-rate growing 82%. "I think they're going to follow Microsoft and be a little bit more mindful of free cash flow, but we'll see." Channel checks put its 2028 TPU target at 12-15 million units (V9, four compute dies) — unreachable with TSMC alone.
Alphabet raised its 2026 data-center budget again, to $195-205 billion, and had a quarter with negative spare cash. That's why it sits in the group dragging the index down. But Singh puts it on the disciplined side of the divide with Microsoft — both are still generating positive cash flow overall, and Google's cloud business is growing 82% at a $99 billion annual pace.
The interesting detail is its own chips: industry checks suggest Google wants 12-15 million of its in-house TPU processors deployed by 2028, more than doubling its needs. That is more than its main manufacturer TSMC can supply alone — meaning Intel has to succeed at advanced chip packaging for Google's plan to work. He expects Google to follow Microsoft in becoming "a little bit more mindful of free cash flow."
Full passage: premium transcript (PDF).
In short: Berkshire's ~$30B position was Buffett's own call, not Abel's — he's had "a front row seat on Google for a long time" via GEICO's ad spend (20 bucks a click costing Google under a cent) and a pre-IPO visit from Sergey and Larry. But Pabrai tempers it twice: Buffett only makes 4-trillion-cap bets because his universe is tiny ("if he were managing 100 billion, I don't think there'd be a Google bet"), and Google is "now a company with debt" and high capex for the first time.
Berkshire put roughly $30 billion into Alphabet, and Pabrai says it makes far more sense that Buffett made that call than his successor. Buffett has watched Google's economics from the inside for two decades: Berkshire's car insurer GEICO buys Google ads, paying about $20 for a click that costs Google less than a cent to deliver — spectacular profitability he could see up close and long regretted not acting on. Google's founders even came to him for advice before the company went public.
But Pabrai is careful not to read it as a recommendation for ordinary investors. Buffett is managing so much money that only trillion-dollar companies are big enough to matter to him — "if he were managing 100 billion, I don't think there'd be a Google bet." Buffett himself downplayed it, saying he doesn't like it as much as four or five businesses Berkshire already owns.
And the company itself has changed shape. Alphabet has never carried debt before, and never had to spend heavily on physical assets; now it has both, because the AI arms race forces it to. That's a materially different, more capital-hungry business than the one Pabrai admired — hence a neutral, observer's stance rather than a buy.
8:44— He was almost like downplaying it. — So, Google is now a company with debt. — Sure, yeah. — They've never been a company with debt. They've never been a company with high capex. Now they have high capex. And all these guys have high capex. In fact, Zuckerberg came out and said that whether the bet works or not, we don't know.
In short: Third in the cloud share table at 15% (behind AWS's 28% and Azure's 20%) and growing faster than AWS off a smaller base — part of the 63% of cloud infrastructure the big three now control. More pointedly, Alphabet is the other name where the buildout has broken the cash engine: hyperscaler TTM operating cash flow grew 34% to $660B, yet free cash flow has turned negative at both Amazon and Google, and Alphabet's funding options explicitly include equity issuance — the sharpest version of the newsletter's recurring cash-vs-debt-vs-equity signal. A disclosed author holding. (Referenced; not a stance call.)
Alphabet appears twice here. First as Amazon's competitor: Google Cloud holds 15% of the cloud-infrastructure market versus AWS's 28% and Microsoft's 20%, and it's growing faster than AWS — though from a much smaller base, and every provider is currently limited by how fast it can build, not by demand.
Second, and more interestingly, as the clearest case of the AI buildout outrunning the cash engine. The big cloud companies together still generated 34% more operating cash than a year ago ($660 billion), but they are spending more than that: free cash flow has gone negative at both Amazon and Alphabet, and is near zero at Meta. The next phase involves borrowing — and in Alphabet's case, the funding options explicitly include issuing new shares, which dilutes existing owners. For a company with one of the strongest balance sheets in the world to reach for equity is the single loudest signal in this whole capex cycle. Referenced, not a stance call.
In short: The consumer winner and, on his read, the overall one: "who's the winner in search? That's just Google… I think Google wins in consumer. They have the complete stack. I think they win in AI overall." Also the exhibit for where the capital is going: the physical infrastructure spend is "where all of this money is being spent, including Google that raised $85 billion recently."
Niles' starting point is a counting question, not a valuation. Look at how many lasting winners a platform category has historically produced: search produced one (Google), e-commerce produced one (Amazon). So he refuses the idea that "five different guys" can win in AI.
Google is his consumer winner, and the reason is that it owns the complete stack — its own chips, its own data centres, its own models, and the products hundreds of millions of people already open every day. Everyone else has to rent at least one of those layers from somebody, which means paying a margin to a supplier and depending on a partner. Owning the whole chain means the profit stays inside the company and no link can be squeezed by a rival.
He goes further than "wins consumer": "I think they win in AI overall." And Google appears a second time as the evidence for where the money is actually being spent — the $85 billion it raised recently is, in his framing, a marker of the scale of the physical build-out, not a warning sign.
0:47If you think about it, you go, well, who's the winner in search? That's just Google. What about in e-commerce? That's just Amazon. Are you going to have five different guys win in AI? No. I think Google wins in consumer. They have the complete stack. I think they win in AI overall. But then in corporate, you have Anthropic, which as I said earlier, they got to profitability in Q2 and their revenues are ramping like nothing we've ever seen in history for a company of that size.
In short: Named in the same 49%-Information-Technology sidebar challenge and inside the Mag 7 group priced at 150x run-rate free cash flow (two of the seven with negative FCF, one at 700x, generously averaged at 100x). One of the AI data-center hyperscalers covered by the roundtable whose conclusion, per Fredrik Tjernstrom's May letter, was "essentially a sell recommendation."
Full passage: premium transcript (PDF).
In short: His biggest position (~$200,000 across both books, ~$110,000 in gains) and the template for the whole thesis. He argued it for over a year "on the back of its earnings reports"; quarter after quarter "all of the bears slowly got their concerns dismissed and beaten down by the data," and the market re-rated it "from a 15 PE to a 25" — the stock up over 100%. "Google was the first of these capex spenders to have its value realized by the market."
Google is his largest holding — about $200,000 across both portfolios with roughly $110,000 of gains — and it's also the proof of the pattern he's betting on with the other three. For over a year he argued the case purely from the earnings reports: every quarter the numbers were excellent, and every quarter one more bear objection got "dismissed and beaten down by the data."
Eventually sentiment flipped from "the capex spending isn't so bad" to "the capex spending is good," and the market re-priced the stock from 15 times earnings to 25 — a doubling. That's what a re-rating is: the business didn't suddenly change, the price investors were willing to pay for each dollar of profit did. Google was "the first of these capex spenders to have its value realized by the market," and he expects Microsoft, Amazon and eventually Meta to walk the same road.
4:57It was all in the numbers. Google was just an incredible company that the market was undervaluing. And as we saw the earnings reports get released quarter after quarter where they put up astounding numbers where all of the bears slowly got their concerns dismissed and beaten down by the data, we saw Google's stock starting to rise.
In short: Split down the middle by the new framework. As a hyperscaler: "the hyperscalers like Google, Amazon, Microsoft, and Oracle have real businesses here. What the returns will look like, I don't know yet, but they have real businesses." As an LLM provider: "partially Google and Microsoft are much more problematic… their LLM businesses are also questionable" — though both are cushioned by "multiple revenue streams from established businesses which are very unlikely to simply disappear." Also the credit-market benchmark: the borrower "they know can pay the money back."
Google appears on both sides of the new dividing line. As a hyperscaler — a builder and operator of the giant data centres everyone else's AI has to run inside — it is one of the four with "real businesses here," protected by the fact that "the amount of money it takes to be a hyperscaler is insane. And that expenditure itself is a moat."
As an LLM provider — a maker of AI models sold to customers — it is in the vulnerable group with Anthropic and OpenAI, facing cheap Chinese open-source alternatives and customers who switch between models to cut costs. The difference is that Google (like Microsoft) has "multiple revenue streams from established businesses which are very unlikely to simply disappear," so a weak model business is survivable rather than fatal.
One more role: Google is his benchmark borrower in the credit market — the company lenders "know can pay the money back," against which CoreWeave's 9% cost of debt is measured.
6:14There are only going to be a few hyperscalers. So the hyperscalers like Google, Amazon, Microsoft, and Oracle have real businesses here. What the returns will look like, I don't know yet, but they have real businesses. The large LLM providers, Anthropic and Open AI and partially Google and Microsoft are much more problematic.
In short: The benchmark that keeps Microsoft's reacceleration in perspective: Azure's 43% "still trailed Google Cloud's extraordinary 82% segment growth." On policy, Google was initially absent from the open-weight letter but Sundar Pichai later endorsed the effort on the company's behalf — leaving Anthropic as the sole major holdout. A disclosed author holding. (Referenced; not a stance call.)
In short: He sold it: "I sold my Google a couple of months ago. I've owned Google. I can't even tell you how long I've owned Google, but I felt I wanted to reduce my exposure to AI." A lightening, not a short — and he has not redeployed: "I haven't bought anything to replace it… I'm just sitting. I've got cash."
This is the news in the segment: Eisman sold Google a couple of months ago, a stock he had owned so long he "can't even tell you how long." He didn't sell because Google is a bad company — he sold "because I felt I wanted to reduce my exposure to AI." Google is one of the biggest spenders on AI data centres, so owning it is one of the purest ways to be long that bet.
Two details make it a lightening rather than a call for collapse. First, he will not short the market — he simply owns less of it. Second, he did not put the money into anything else. He is "just sitting" in cash, because in his view the usual defensive swap doesn't work here: "buying something defensive or stapley is kind of a waste of time… they don't go up. People either want to buy AI or they don't want to buy AI, but they don't want to shift out of it to buy Clorox." When one trade dominates a market, money that leaves it tends to leave the market altogether rather than rotate into safer stocks — so cash, not staples, is the exit.
0:23I've owned Google. I can't even tell you how long I've owned Google, but I felt I wanted to reduce my exposure to AI. I haven't bought anything to replace it because buying something defensive or stapley is kind of a waste of time. Why, that doesn't go up. They don't go up. They — people either want to buy AI or they don't want to buy AI, but they don't want to shift out of it to buy Clorox.
In short: Luria: "Google is an AI winner" — a year ago at $180 it was the AI loser; by end-2025 it was the only winner ("a flippity flip"); now there's a pullback because "Google's model is no longer state-of-the-art… a distant second consumer chat… distant third on enterprise AI. So maybe they're not the winner, they're a winner." The kicker: search advertising growth accelerated — "instead of it dying, it accelerated… because they're using AI to sell us more ads for more money." Ives: "the best positioned hyperscaler relative to the crowd," and the $85B raise "was the right move."
Google's AI reputation has swung completely twice in eighteen months — from certain loser (AI chatbots kill search, regulators break it up) to sole winner (its own chips, its own model, its own cloud) and now to something in between. Luria's landing point: "maybe they're not the winner, they're a winner."
The negatives are real: the Gemini model is "no longer state-of-the-art," it's a distant second in consumer chatbots and third in selling AI to enterprises, and a bureaucracy loses researchers to startups. The positive is the part everyone predicted would break: search advertising growth accelerated, because AI lets Google sell more ads for more money. Cloud growth accelerated too. Ives calls it the best-positioned hyperscaler and defends the $85 billion share sale as the right move — in an arms race, being unable to spend is the bigger risk.
20:59Yeah. What are your thoughts on Google in this world? The quick thought is Google is an AI winner, right? I think a year ago when the stock was at 180, everybody considered them the AI loser because search is going to go away. We're not going to use search anymore. Break them up.
In short: Still "my favorite in the space for a long long time," but the earnings quality is the warning: of the very good results reported that day, "almost two-thirds… have come from the gains on their venture capital investments in these entities which are big customers" — booked in other income, legitimate but "very unsustainable" — "it is a circular financing type of situation." The stock got "a whack" despite good numbers because it raised capex targets, and a large depreciation schedule has yet to hit the P&L. Also the broader lament: these were capital-light, high-FCF businesses once at "10 11 times earnings" — "maybe someday they'll be back there given what they're doing."
Google's parent has been "my favorite in the space for a long long time," so this is a caution about the numbers rather than a rejection of the company. The catch is where the profits are coming from: of the very good quarter reported the day they spoke, "almost two-thirds" of earnings came from writing up the value of Alphabet's stakes in AI companies — which are themselves big paying customers of Alphabet's cloud.
That is what he means by "circular financing": Alphabet invests in an AI firm, the AI firm spends the money buying Alphabet's computing, and Alphabet books both the revenue and a paper gain on the stake. It is entirely legal — accounting rules require those gains to be recorded — but it isn't repeatable, so headline earnings flatter the business.
There is a second, delayed problem: the enormous spending on data centres is written off gradually as "depreciation," a cost that "has really yet to hit the P&L." Strip out the one-time gains and add the depreciation coming, and a stock that looks like it trades at a teens-to-twenties multiple of earnings is far more expensive. The market seems to be noticing — Alphabet fell on good results because it raised its spending plans, which is exactly what happened to great companies at the top in early 2000.
38:59up here in a bit we saw with uranium so it's like the whole complex got sold off very very hard and I think created just a tremendous buying opportunity. All right, let's skip to Oh, there's your AI. — Yeah, — this is So, this gets to what you're saying and it's pretty amazing like with Google's most recent earnings which were very good almost twothirds of them have come from the gains on their basically their venture capital investments in these entities which are big customers. So, it is a circular
In short: The deep-dive and the template for the week: record $112B net income of which 87% was markups on SpaceX ($94B stake, $80B sale-restricted) and Anthropic — ex-markups net income grew only ~2% sequentially, hence the 6% drop. First negative FCF quarter since going public (−$5.9B); capex guided up to $195-205B (Q2 $45B vs $44B est); purchase commitments $811B, up $500B in three months. Cloud $27.5B (+82%) with backlog $514B, but at lower margin; Search $63B a slight miss; YouTube $11B a beat; Gemini 950M MAUs (below Q1). Plus the advertiser teardown — deprecated second-price auction, diluted exact match, budget caps blown by 2× — "the actions of a company whose core business is in decline… at the altar of AI capex, Google is sacrificing its golden goose of search." Not a sell call, but the hyperscalers are a wait.
Alphabet's headline quarter was a record — $112 billion of net income — but 87% of it wasn't from operating the business. It came from marking up the value of its stakes in SpaceX and Anthropic on paper (and it can't even sell $80 billion of the SpaceX stake yet). Strip that out and profits grew about 2% from the prior quarter. That's why the stock fell 6%.
Underneath is the real story: for the first time since it went public over 20 years ago, Alphabet burned cash — free cash flow of minus $5.9 billion — because it is spending $195-205 billion this year on AI infrastructure. And its contracted future commitments (chips, data centres, electricity) jumped to $811 billion, up half a trillion in three months. The cloud business is genuinely booming (+82%, with a $514 billion backlog), but at lower margins.
The uncomfortable part is the advertising teardown Singh relays from a large advertiser: Google quietly abandoned its long-standing "second-price auction" (where you paid a penny more than the next-highest bidder, not your own maximum bid), widened keyword matching so advertisers pay for irrelevant clicks they can't switch off, and started blowing through daily budget caps by 2×. That's how you grow revenue while search volume shrinks to AI chatbots — and it's borrowing from the future. His verdict: "at the altar of AI capex, Google is sacrificing its golden goose of search." Not a short, but wait.
Full passage: premium transcript (PDF).
In short: The autonomy yardstick that frames how early Tesla still is: Waymo has accumulated roughly 220 million autonomous miles against Tesla's 380K unsupervised miles across six cities (Robotaxi live in seven metros, weekly mileage growing double digits, Cybercab now in production accumulating calibration miles). Also the AV competitor implicitly behind Uber's "AV hedge" rationale for buying delivery scale. A disclosed author holding. (Referenced; not a stance call.)
In short: "Very mixed results at best." He strips out the investment gains Google books in EPS: adjusted EPS 285 vs 231, a miss vs 289; revenue +24% and Cloud $24.8B "up an incredible 82%." The problem: "because of the massive AI CapEx spending, free cash flow turned negative. It was negative 5.9 billion," and 2026 AI capex went $190B→$205B. "The market is beginning to lose patience with all this crazy spending" — down 7% Thursday, dragging Nasdaq >2%.
Two things Eisman does before judging the quarter. First, he strips out Google's gains on its investment stakes, which flow through reported earnings but aren't the operating business — on that adjusted basis earnings missed. Second, he looks past revenue (which was strong: +24%, with cloud +82%) to cash.
And the cash is the story: after all the AI spending, free cash flow turned negative — minus $5.9 billion. Free cash flow is what's left after running the business and paying for equipment and buildings; when it goes negative, growth is being funded from the balance sheet rather than from profits. Google also raised its 2026 AI budget again, from $190 billion to $205 billion. A year ago that raise would have been cheered; this time "the market is beginning to lose patience," and the stock fell 7%.
11:23But, revenue increased an impressive 24%. Google Cloud revenue reached 24.8 billion, up an incredible 82%. So, so far, not bad. On the bad side, however, because of the massive AI CapEx spending, free cash flow turned negative. It was negative 5.9 billion. Moreover, Google upped its 2026 AI CapEx spend from 190 billion to 205 billion.
In short: The earnings-quality tell: Google "puts up beats and raises capex, but there's that pesky FCF -$6bn," and 87% of all its 2Q "income" was "Other" from marking up its own stakes in Anthropic and other privates — the "Great Circularity in the digital god ponzi." Framed as evidence the hyperscaler "front end" is where FCF is plummeting (the Cembalest train).
Google reported a "beat" and raised its spending plans, which normally sends a stock up. Paulo's point is that the quality of that profit is rotten. Two things: first, its free cash flow — the actual cash left after running the business and building data centres — was negative $6 billion (a company this size gushing cash is now burning it, because the AI build-out is so expensive). Second, and worse, 87% of what Google called "income" last quarter wasn't from selling ads or cloud — it was "Other Income" from marking up the value of its own stakes in private AI companies like Anthropic.
That's the "Great Circularity" he keeps flagging: Google invests in Anthropic, Anthropic commits to spend on Google's cloud, and each new fundraising round lets Google write its stake higher and book that paper gain as "earnings." No cash changes hands, but it flatters the profit line — the private-equity trick of marking up your own holdings, now run through a public company. So he reads this "beat" as evidence the hyperscalers (the front of the cycle) are where cash flow is actually cratering — the opposite of a healthy result.
In short: Same role as Meta — a hyperscaler whose data gravity advertisers are wary of, which is why they'd rather have Zeta marry their own first-party data to Zeta's. Reference only.
55:21So really they have been very good at optimizing companies advertising budgets and figuring out there's a lot of companies right now that don't necessarily trust the hyperscalers like the Metas and the Google's with all their proprietary data. So Zeta comes in and they show you how to optimize your proprietary data, marry it with their data, right? And then use agents and new AI tools and again it's got clean balance sheet.
In short: Per Niles, Google is his favorite consumer-AI play (a complete AI stack; GCP growth accelerating 34%→63%) and reports Wednesday — the first big token-economics data point. But it's not immune to the top-1% cutting AI bills, and "the market's going to focus very closely on margins."
Via Dan Niles: Google is his favorite "consumer AI" bet because it owns the whole AI stack (chips, cloud, models, products), and its cloud growth has been accelerating sharply. It reports Wednesday and will be the first real read on whether cheaper AI tokens help or hurt. The caution: even Google isn't immune if its biggest customers cut their AI bills, and the market will zero in on profit margins. Presented as a key data point, not a table-pounding buy.
Full passage: premium transcript (PDF).
In short: The share winner in the attention fight: YouTube lifted its slice of US TV time to a record 13.4% in April — roughly 1.7x Netflix's 7.8% — inside a streaming category that reached 47.6% of viewing (from 44.3%) as cable fell from 24.5% to 21.6%. The pattern the article draws: the platforms gaining the most attention don't rely on expensive scripted series, they combine creator content, sports, free programming or multiple entertainment formats. (Referenced; not a stance call.)
In short: Raised ~$80B, upsized to $85B; struck a short-dated "spot" compute deal with xAI (with "tremendous outs"); a hyperscaler in the ROIC-compression group.
Google raised about $80B (upsized to $85B) and signed a short-term deal to rent computing power to Elon Musk's xAI — a deal with "tremendous outs" (either side can walk within months). Chanos reads these as promotional, get-it-on-the-tape arrangements rather than durable demand.
Like the other hyperscalers, it's profitable but caught in the same falling-return-on-new-spending trend.
7:12When you think about issuance, right? So SpaceX does a $75 billion IPO, and then they follow on with $25 billion in debt almost immediately. This is a week after Google, and I know you've had a lot to say about that, raised what was that number? 80 billion, and then it was upsized to 85, right? And then we know that Meta, and the list goes on and on.
In short: Named with Amazon as a hyperscaler moving directly into nuclear power to feed data centers — the first choke point before fuel supply becomes the next one.
0:58When Microsoft went to the utility that owns it and said, "We want to contract this much power out for this many years," all of a sudden it became economic to do that again. But we are seeing the Amazons of the world, the Googles of the world actually getting right into it because we're now seeing all of these data centers requiring immense amounts of electricity and power that doesn't exist right now.
In short: The day's story: Buffett told Becky Quick he initiated Berkshire's Alphabet stake, called not buying Google earlier "a mistake," and said the company is now "more likely to be a winner based on its record." Sechan: "that's our largest overweight in the mega caps" — some of the most attractive characteristics; a year ago it was "indicted as being taken over by chat and Anthropic," and that fear created the opportunity in a well-capitalized name.
Alphabet is Google's parent. The day's catalyst: Warren Buffett revealed that Berkshire Hathaway has bought an Alphabet stake, admitted that not buying Google years ago was "a mistake," and said the company is now "more likely to be a winner." Rob Sechan says Alphabet is his firm's "largest overweight in the mega caps." His point about opportunity: just a year ago investors feared chatbots (OpenAI, Anthropic) would destroy Google's search business — that fear knocked the stock down and created the buying opportunity in a company with huge financial resources.
In short: A survivor with real cash flow (search, YouTube) to fall back on — one of only two hyperscalers currently cash-flow positive. But depreciation expenses double (17%→35% by 2028), so the PE "adjusted for non-success" is much higher than the ~24× headline.
Google's saving grace is that its old business — search and YouTube ads — still gushes cash, so it has something real to fall back on when the AI spending stops paying off (it's one of only two hyperscalers still cash-flow positive after data-center costs). The catch is accounting: the huge sums spent on chips and buildings get expensed slowly over years ("depreciation"), and Google's depreciation roughly doubles by 2028 (from 17% to 35% of something). If AI revenue doesn't show up to cover it, real earnings sink — so the stock's "reasonable" ~24× earnings is much richer once you assume the buildout disappoints.
20:01And then only now the only ones that are positive cash flows are Google and Microsoft. The other ones all cash-flow negative. How they expect by 2028, 29 and 2030 those cash flows to soar, their free cash flows, but also their earnings. And the problem is, with the depreciation expenses going up over time, if the revenues don't flow, then, for example, Google's depreciation expenses are going to rise from 17% up to 35% by 2028.
In short: In the motivation framework Google is "in between — partially commoditized, partially premium": it shares Meta's distribution + billions of users and wants low AI unit cost, but also keeps a premium, charged-for model. Grouped with Meta/Amazon as a commoditizer that wins the tug-of-war over time. Also "a massive winner in the portfolio along with Google for the past year" (named alongside ASML as what's working).
In Carlson's framework Google sits in the middle — "partially commoditized, partially premium." Like Meta it has huge distribution (billions of users across Search, Gmail, Maps, Android, YouTube) and wants the cost of running AI to be very low, so most of its money comes from that distribution and its own cloud. But unlike Meta it also keeps a premium, paid-for model on the side.
He lumps Google with Meta and Amazon as one of the "commoditizers" he expects to win the AI tug-of-war, and separately notes Google (alongside ASML) has been "a massive winner in the portfolio for the past year." So it's both a framework winner and a proven holding.
15:37That's Meta's goal. So Meta has the motivation of highly commoditized. Google I think is in between. I put them as partially commoditized, partially premium. So Google shares a lot of the attributes of Meta. They have the distribution layer, the billions of users, the tons of apps to monetize from.
In short: Passing earnings-quality contrast (negative-lean, not a call) — named as an example of S&P "heavy hitters… recording massive non-recurring gains on investments in private companies (many of which are also their customers)," inflating index earnings; LMT's 17× P/E, by contrast, isn't propped by extraordinary items. Cited to frame LMT's relative cheapness, not an argued view on Alphabet.
In short: Ellenbogen: the only publicly traded LLM (Gemini) — the direct proxy for LLM-layer investment; stock has had a strong year; its TPUs are among the low-cost ASICs Anthropic built on. Giroux: hyperscaler FCF winner 2030–32. Black: 25x, reasonably priced.
In short: Its Gemini Enterprise Agent Platform (via Model Garden) already offers a broad selection of Chinese AI models (DeepSeek, MiniMax, Moonshot, GLM, Qwen) fully managed inside the US cloud ecosystem — the open-source distribution channel underpinning GS's "going global" leg.
Full passage: premium transcript (PDF).
In short: Guided 2026 AI capex from $90B (2025) to $180B, and "in June, Google raised 85 billion in equity capital" — a company that "historically… don't ever raise capital" now with a "massive need." The tell that software has turned capital-intensive.
Google is Eisman's cleanest example of AI's "capital-intensity sea change." Its AI capital spending is guided to roughly double — from $90 billion in 2025 to $180 billion in 2026 — and in June it raised $85 billion of new equity (selling stock). Companies like Google historically never raise money; they throw off so much cash they struggle to spend it. Needing to raise capital signals the business has quietly turned into a heavy, hardware-hungry one.
That's the worry for the whole group: shareholders are now being asked to fund a build-out whose returns aren't proven, in a field Eisman thinks has no lasting competitive moat.
10:54Take Google as an example. In 2025, Google spent 90 billion on AI capex. When it reported fourth quarter numbers, it guided 2026 AI capex to 180 billion. Wow. Meta guided to 135 billion and Amazon guided to 200 plus billion. These numbers started to scare people. But what really freaked them out was that in June, Google raised 85 billion in equity capital.
In short: Brown's "double benefit" name: Alphabet both spends the CapEx and owns the LLM — "Gemini will profit heavily from the expense that Alphabet the parent is plowing in." Value accrues on both ends, even if the picture is murkier than Apple's.
Alphabet (Google's parent) is unusual in this debate because it plays both sides: it's a big AI spender, but it also owns a leading AI model, Gemini. Brown calls this the "double benefit" — the money Alphabet spends building AI infrastructure comes back through Gemini's own profits, so value accrues on both ends rather than leaking out to suppliers.
He's less certain than he is on Apple — the enterprise-AI picture is "murkier and harder to figure out" — but Alphabet is on his short list of companies positioned to actually capture the payoff from the AI build-out.
In short: A held winner, not a fresh July buy — "I have a couple companies that have done well in this market. Google's one of them… continues to go up any day the semiconductor trade goes up." Also his lead operating-leverage example: adjusted EBIT/employee rising from $393k (2022) toward projections through 2028 while headcount stays flat.
28:21In the meantime, I have a couple companies that have done well in this market. Google's one of them and ASML is another one that continues to go up any day the semiconductor trade goes up. But I'm building up the rest of my portfolio for when this trade happens. Now, as we jump into some news here, we have Tom Lee explaining why he believes this month will be particularly good in the market.
In short: Disclosure-only — one of the App Economy author's portfolio holdings (owns GOOG); not discussed in the Cerebras body. Referenced/neutral.
In short: His best past pick (doubled); still owns. Bought at 18x when everyone said AI would kill search — AI, Cloud, YouTube and Waymo all compounding. Only risk is the heavy AI capex / first equity issue in 15-20 years.
Alphabet (Google) was his best past pick — it doubled. He bought it when it traded at 18x earnings (cheaper than the market) and everyone assumed AI would destroy Google search. Instead AI, the cloud business, YouTube and the Waymo robotaxi are all growing fast.
He admits he made the mistake of selling near the peak Gemini-panic, then learned to "never sell" quality. He still owns it. The one risk he can't fully answer: Google is pouring its cash flow into AI and even did a rare big stock sale to fund it — will that heavy spending ever stop, and what happens if it does?
39:02Um, let's talk about Google. It's doubled since then. That's the best one. Up 104% since you chose it. Somehow I'm down on Google. That's That's impossible, but it seems impossible, but that's because I was an idiot and I sold during the peak Gemini fear and that was the exact
In short: His other named hyperscaler in the "lag-7" setup — same thesis as Amazon: margin pressure from chip-input inflation, not a fatal flaw, expectations too high; the pullback is a setup for the next leg higher. (Joining the Dow; Google Cloud grew 63%, ~24×, per Link.)
In short: Alphabet at 24x with Google Cloud growing 63% — the fastest cloud grower of the three and the cheapest-looking against it. Joining the Dow. Attractive on the multiple-vs-growth screen.
Alphabet is the cheapest-looking of the three on Link's growth-vs-price test: about 24 times earnings while Google Cloud grew 63% — the fastest cloud grower of the bunch. It's also being added to the Dow. Her read is that you're paying a similar multiple to Amazon for noticeably faster cloud growth, with the same caveat about the cash being plowed into the AI build-out.
In short: "The best chart of the big names" — got very extended/super overbought, so take some money off the table and revisit, but it stays viable. The big-cap most like GEV (the power inside Google's AI).
Among the mega-cap charts Sohn flips through, Google's is "the best of the big names." It has gotten very stretched (super overbought), so his discipline is to take some money off the table and revisit — but the underlying trend stays healthy and the stock remains viable.
He links Google to GE Vernova because of the "power in the AI of Google" — i.e. it's on the right side of the one theme driving everything (compute and the power behind it), unlike Microsoft, Meta or Oracle, whose charts he finds weak or messy.
10:34GOOGL (Google/Alphabet) — different story. Got very extended, super overbought — take some money off the table and revisit; it will likely be viable. Compared on one screen, Google has the best chart of the big names — most similar to GEV (power in the AI of Google).
In short: "Interesting long-term," add on the selloff but not aggressively: the strongest balance sheet (negative net debt, a 100-yr bond, $85B equity raise) funding $180-190B 2026 / $200B+ 2027 capex; FCF collapsed but Cloud +63% to ~$20B at 32.9% margins with a ~$168B backlog. The capex keeps climbing, so size small.
Full passage: premium transcript (PDF).
In short: His largest position; organic (Ebit/operating) growth ~20–21% next year once you strip last year's one-time equity gains. Moat "clearly gotten stronger" — no chance of losing Chrome, cloud now a bigger/Moody-er share, and Waymo has moved from project to "a real business now." Disagrees with the AI model's cautious "interface risk" read; cites Chris Hohn (the "moat guy") buying in.
Google is Carlson's biggest holding, and he thinks it's getting stronger, not weaker, despite fears that AI chatbots will eat search. His reasons: Google has locked in Chrome, its cloud business is growing into a bigger and more reliable ("Moody-er") slice of revenue, and Waymo has graduated from a science project into a real, money-making business. Stripping out a one-time accounting boost from assets that rose in value, the underlying business is growing around 20–21% a year. He notes even Chris Hohn — a famously moat-obsessed investor — has bought in.
In short: The "no moats / no non-competes" exhibit: its Gemini co-head left for OpenAI and a senior DeepMind engineer left for Anthropic — "if engineers can switch companies at will, product differentiation becomes almost impossible." (Communication services were +7% in Q2 "but much of that performance was just from Google being up 20%.")
Google's negative news this week wasn't the business — it was people walking out the door. The co-head of its Gemini AI models left to join OpenAI, and a senior engineer from its DeepMind lab left for Anthropic. Eisman's point is structural: if the best engineers can freely jump from one AI company to the next (no non-compete agreements stopping them), then no company can keep a lasting technical edge — the "no moats" problem made concrete.
That talent flight is why he frames the news negatively even though Google's stock was actually up 20% for the quarter (it carried almost the whole communication-services sector). The worry is about durability, not this quarter's price.
6:50These negative changes to the hyperscaler story were partially driven home on Monday when it was reported that two Google senior officers left to go to OpenAI and Anthropic. Google's co-head of its Gemini AI models left to go to OpenAI and a senior engineer in Google's DeepMind left to go to Anthropic. Good news from Micron. Micron reported and their numbers were incredibly powerful.
In short: Passing reference — cited as a competitor in AI coding, and (via YouTube) the default video platform pressuring the Fox/Roku living-room thesis.
In short: Defends it on a −6% "brain-drain" day (Noam Shazeer → OpenAI, John Jumper → Anthropic). Investors "over-index on a few top talents"; the real AI moat is distribution + the full stack (TPUs, cloud, billions of users across Gmail/Maps/Android/YouTube), not two or three researchers.
Alphabet (Google) fell 6% in a day after two star AI researchers quit — Noam Shazeer left for OpenAI and John Jumper left for Anthropic — and the market shaved tens of billions off the company over the "brain drain." Carlson thinks that's a mistake. He argues investors wildly over-rate how much a couple of brilliant individuals matter to who wins in AI.
His point: the durable advantage in AI isn't having the single best model or smartest people — it's distribution and owning the whole stack. Google has its own chips (TPUs), its own cloud, and AI built into products billions of people already use every day (Gmail, Maps, Drive, Android, YouTube, the Gemini app). That end-to-end control is the moat, and it doesn't walk out the door when two or three researchers leave. He expects future earnings to keep proving that the infrastructure — not any one genius — is what makes Google win.
25:23Now, the first bit of news that we get into today is that Google stock is going down. It's down 6% on the day. Huge red day. So, what's going on with Google? Well, the big news over the weekend is that Google has faced its worst day because of AI concerns after high-profile exits.
In short: Named among the big names "there or thereabouts" at the Silver Springs land — the data-center cluster anchoring Comstock's real-estate value. (Also Waymo's parent.)
In short: The one big-tech name where he concedes value ("you can say that about Google"), unlike Nvidia — millions of revenue counterparties, not a handful. But Alphabet's equity raise is part of the funding-window risk.
10:05That only, again, I'm going to push you on this because that really, that happens because the growth stocks are expensive. Right. Those growth stocks, you could argue, are now value stocks. Okay. I mean, you can't say that about NVIDIA. You can say that about Google in respect. Yeah. And you can say Seth Klarman, the most famous value investor.
In short: DOWNGRADED BUY → HOLD — "Technology and internet services company." No figures are published for it in this issue, and no reason is given beyond the rating change. Note the timing: four days after the superinvestor issue reported Chris Hohn making Alphabet 3x his Microsoft position, the archive moves the other way on both names. It is upgraded back to Buy in August.
In short: The capital-intensity exhibit: raising $80B (upsized to $85B), all equity — first equity raise since June 2005. 2025 AI capex ~$80B (mostly cash-flow funded) → 2026 $180–190B, "just too much for its cash flow." Software has become a capital-intensive hardware business.
Google is Eisman's headline example of AI's "sea change." Software companies normally throw off so much cash they almost never sell new stock — Google hadn't raised equity since 2005. Now it's raising up to $85 billion, all in new shares, because its spending on physical AI gear jumps from ~$80 billion last year to $180–190 billion this year, more than even Google's cash flow can cover.
The point isn't that Google is doomed; it's that a cash-machine software business has quietly become a capital-hungry hardware business — and shareholders, not just internal cash, now have to fund the build.
7:54In 2026, Google will be spending 180 to 190 billion on AI capex and that is just too much for its cash flow. There are also stories that Meta and Microsoft will be doing similar transactions soon. Oracle increased its capital raising plans by 20 billion recently. This all goes to show that certain non-capital-intensive large software companies have now become capital intensive hardware companies.
In short: The bellwether of tech over-investment: "free cash flow yields are likely even at Google are likely to go negative… they're investing too much," exactly the 120%-of-cash-flow signal the oil producers flashed at the 2014 top. "Which one's going to have a rising ROC? The oil guys, not the tech guys."
Currie uses Google as the poster child for tech spending too much. "Free cash flow" is the cash a company has left after it pays for running and growing the business; the "yield" is that cash measured against the stock's value. He thinks Google's free-cash-flow yield is heading toward negative — meaning the AI build-out (data centers, chips) is now eating more cash than the business throws off.
That matters because it's the exact warning sign the oil producers flashed at the 2014 top, when they were plowing 120% of their cash flow back into the ground and the sector then collapsed. He's drawing the parallel in reverse: the AI giants have turned into capital-hungry, cyclical "commodity" businesses (an AI data center is like an oil refinery — inputs in, a commodity called "compute" out) yet still trade at 30–40× earnings, the multiple of an asset-light growth company. His bottom line: "Which one's going to have a rising return on capital? The oil guys, not the tech guys."
26:25There was no free cash flow left over. 120% of earnings. And we're getting that close on the tech side. They're burning through the cash. Free cash flow yields are likely even at Google are likely to go negative. Which is telling you they're investing too much. And so you look at the free cash flow yields in the energy guys, they're — Mhm.
In short: The "most striking" culture-FUD-then-recovered example — he "was in this stock" when the culture was mocked at ~$70 (2022), Chris Hohn called staff overpaid; now ~$360 and nobody asks about culture. Cited to validate the Meta thesis.
14:03Netflix staffers voiced frustrations and fears of cutbacks ahead. Problems internally with Netflix. The company's disjointed. It's falling apart. And look at the stock price. Of course, it is. You don't hear the same issues about Netflix's culture problems today because the stock price has recovered. Again, this pseudoscientific, backwards-looking, hindsight investing did not work with Netflix in the past, and it won't work with Meta today.
In short: Owns a frontier model (Gemini) but inside a business with Search, Android and Cloud to absorb a shutdown — the cushioned end of the "owning the frontier" spectrum. Also the winner of Apple's ~$1B/yr Siri deal; Gemini runs on NVIDIA chips in Google's data centers.
Google builds its own frontier AI model (Gemini), which normally looks like the riskiest position to be in — if a government forces a model offline, you lose your crown jewel. But Google is cushioned: Gemini sits inside a company that also runs Search, Android and Google Cloud, so a model shutdown wouldn't take the whole business down with it. That's the opposite of a pure-play AI lab.
Google is also quietly winning from everyone else's caution: it's collecting ~$1 billion a year to power Apple's Siri with a custom Gemini, and that model runs on NVIDIA chips inside Google's own data centers. So Google both owns the frontier and rents it out — getting upside from the buildout while having other businesses to absorb a regulatory shock.
In short: "Google just sold 80 billion of stock. They're selling bonds. They're doing everything to fund artificial intelligence" — part of the same equity/debt supply wall that makes the megacaps "the most expensive ever" while their free cash flow is consumed.
Same point, a different instrument: "Google just sold 80 billion of stock." Selling shares means issuing new ones — existing owners get diluted — and it lands on top of the largest IPO calendar in history. His conclusion isn't that Alphabet is a bad business but that the supply of new equity, combined with record valuations, makes a flat decade for the index likely.
14:38Look at Nvidia today, a 20 billion-dollar bond offering. The Mag 7 has done almost 300 billion of bond offerings. So, they're selling stock. Google just sold 80 billion of stock. They're selling bonds. They're doing everything to fund artificial intelligence. And so, that creates a dynamic where these companies are the most expensive ever.
In short: Liquidity-drain + circularity in one name: raised ~$85B of equity (one of the largest ever) ahead of the IPO wave, owns 14% of Anthropic, is its primary cloud counterparty, and co-makes the TPUs Anthropic buys with "Big Sky" money.
Full passage: premium transcript (PDF).
In short: The other half of the same disagreement. Hohn "bought Alphabet instead. That makes his Alphabet position 3 times bigger than his Microsoft position" — the reasoning attributed to him: "people don't see how big Google really is, how strong its data is, and how much it still rules search." Ackman went the other way and sold his Alphabet position outright. Alphabet was downgraded Buy → Hold on the 18 June list four days later.
In short: Example of a business with an "indisputable moat," a YouTube/Apple/Amazon incumbent Spotify out-competed, and an organization that "with increasing speed and efficacy" overcame the AI-kills-Google fear a few months ago.
3:14And it stems down to the innovation stack. And the innovation stack can only arise when you have an organizational culture and structure that is obsessively focused on the end user. And in Duolingo's case, I would go as far to say that I don't think I've ever come across a company in my life that is qualitatively as strong as Duolingo.
In short: Raised $85B all-equity — first raise since the 2004 IPO — because AI capex jumps ~$80B (2025) → $180–190B (2026), beyond cash flow. "The most important news of all"; software has become a capital-intensive hardware business. Strong backlog/other reasons to own, but the equity market now funds the build.
Google is the headline event. Software companies historically generate so much cash they rarely need to ask investors for money — Google hadn't raised equity since it went public in 2004. This week it raised $85 billion, all in new stock, because building AI is now staggeringly expensive: its capital spending on physical gear ("capex") jumps from about $80 billion last year to $180–190 billion this year, more than even Google's cash flow can cover.
Eisman calls this "the most important news of all": a cash-machine software business has quietly become a capital-hungry hardware business, and now the stock market — not just internal cash — has to fund the build. He's not condemning Google (there are "other reasons to own" it), but the easy-money software model investors got used to is over.
4:00Google announced that it was raising $80 billion since upsized to 85 billion in new capital all from equity. Historically software has been a non-capital-intensive business. The last time Google raised equity for the company was when it went public in 2004. So why is it doing so? Because the table stakes of participating in AI keep increasing.
In short: Just issued $80B of equity ahead of the IPO wave and is paying SpaceX $920M/month for compute at xAI data centers — both planks of his liquidity-drain and circular-financing concerns.
35:35I'm just not I'm just not the type of person that plays IPOs. Sure. Without some, clever way of getting in cheaper. — By the way, Google is going to pay SpaceX $920 million a month for compute capacity and XAI data centers. my question isn't about this particular deal.
In short: Last week's ~$80B secondary share offering is part of the supply wall — "you got Google dumping stock" alongside Elon and the CFOs, all at the same time.
Google's parent sold roughly $80 billion of new shares last week to fund its AI buildout. In Larry's supply-and-demand framing, that's $80 billion of fresh stock the market must digest at the same moment as the SpaceX/Anthropic/OpenAI IPO wave — "you got Google dumping stock" right alongside Elon and the convertible-issuing CFOs. Too many sellers at once.
7:39So if you do the math, if you take the secondary offering from Google last week of shares, 80 billion plus the SpaceX, you get 150 billion. 150 billion. And so net net a lot of people have to sell stocks in the market to make room for these two absolute beasts that have come out. If you take SpaceX, as you said, it's a $2 trillion valuation.
In short: The most substantive of the group: a top-index constituent passive funds must sell to fund the IPO, and it has already cut its buyback and raised equity ("as we saw from Google") — the dilution moment. Also the illustration of "no one has enough capital": Google is giving a letter of credit to Anthropic. A flow/dilution reference, not a buy.
Smead uses Google to make two connected points, neither of them a buy signal. First, because Google is one of the largest stocks in every index, it's exactly what passive funds have to sell to raise the cash a huge IPO soaks up — so it sits at the center of the "forced selling." Second, and more telling, Google has stopped buying back its own shares and has started raising money instead. When big companies switch from returning cash to shareholders to issuing new stock and debt, that's the point in the cycle where existing owners get diluted. He even notes Google is extending a letter of credit (a funding backstop) to Anthropic — his evidence that in the AI build-out "no one has enough capital." So Google here is a warning flag about where we are in the cycle, not a recommendation.
0:46That's where the forced selling will be the biggest. Um, and what's interesting is those companies in many cases either don't have buybacks or have cut their buybacks and are now raising equity capital, as we saw from Google and what we'll see from Meta. So, I I point that out because uh you know, when the public stock markets are IPOing or offering secondaries like this, that's when the dilution of the capital in the common stock market happens and you wake up years later and you always look back to your point on your career and say, "Gosh, man, I wish I would have
In short: His largest position and "the standout winner," but only 11% off highs and expensive on P/FCF after a big run — he'd wait for a better entry (though he understands Berkshire's $10B buy). Hold.
Alphabet (Google's parent) is his single largest position and what he calls "the standout winner." The catch: it's only 11% off its high after a massive run, and on a price-to-cash-flow basis it's "super expensive."
So while he's not selling, he wouldn't buy more at this price — he'd wait for a better entry point. He does add that he understands why Berkshire Hathaway is buying about $10 billion of it, so he doesn't think buying today is a mistake, just not the best moment.
9:48I don't think that that's going to happen. After Moody's we have Google. The standout winner Google is only 11% from its high which is very normal for a company to trade about 10 to 11% off of their highs. In fact, a lot of companies just naturally trade around that range. So, this is towards the higher end of its 52- week range.
In short: "We love Google… one of our largest positions" — Gemini is the third horse in the foundational-model oligopoly, attached to a huge business.
Google's parent is "one of our largest positions." In the race to build the best core AI models, Sacerdote sees an oligopoly — just three serious players — and Google's Gemini is the third horse alongside Anthropic and OpenAI.
The extra appeal is that Gemini is bolted onto an enormous, cash-rich existing business (search, ads, cloud), which funds the brutally expensive AI buildout. "Gemini can never be counted out."
3:18In the meantime, Anthropic kind of was this dark horse candidate, the startup and um they focused uh really purely on the enterprise and OpenAI had kind of won the consumer and then Gemini can never be counted out. We we love Google as well. It's one of our largest positions. So it really started to look like a three-horse race and somewhat of an oligopoly very similar to how the uh cloud market evolved where three companies underpin the entire SAS cloud world and and have really excellent businesses and then we also were aware of the open- source risk
In short: The headline subject — sold ~$85B of new stock in two days (upsized from $80B) despite $120B+ cash and ~$165B operating cash flow, against ~$190B of 2026 capex. Choosing equity over its strong cash position is the article's central signal about AI-capex strain.
Alphabet owns Google search, YouTube, Android, and a giant cloud-computing business. To build the data centers AI needs, it is spending about $190 billion this year on buildings, chips, and equipment ("capex" = the money a company pours into long-lived physical assets).
The eye-catcher: even though Alphabet is sitting on more than $120 billion of cash and pulls in roughly $165 billion a year of cash from running the business, it still chose to print and sell about $85 billion of brand-new shares in just two days (it had planned $80 billion and raised the amount). Selling shares slices the company into more pieces, so each existing owner's slice shrinks a little. A company this cash-rich doing that is the article's main signal that the AI spending is getting heavy enough to reach for outside money rather than just paying out of pocket. The piece isn't saying buy or sell — it's reading the financing decision as a confidence/strain signal.
In short: A holding; he supports Google's equity raise — at a ~30 forward PE the dilution to get $80B is only ~1.8%, and investors already credit it with decent returns on capital. The viral Gary Marcus "no moat" tweet targets Google, which he rebuts.
Google is the search-and-ads giant that also rents out cloud computing — another of his four AI hyperscalers, and a stock he owns. The viral "AI is just a commodity with no edge" argument was aimed squarely at Google, and he spends most of the video rebutting it.
Unlike with Meta, he's fine with Google raising money by issuing new shares: because Google's stock is expensive (priced at about 30× a year's profits), it only has to give up about 1.8% of the company to raise $80 billion — a cheap way to fund its build-out. Investors also already trust that Google earns solid returns on the money it spends, so the move shouldn't spook them.
20:12And I hope that they don't end up doing this for a couple of reasons. There's two different reasons why I support this with Google and I think it's a good strategy for them, but I don't with Meta. And the first one is valuation. When we look at the Ford PE of Google, it's currently sitting right around 30. So Google's at a 30 Ford PE.
In short: Builds its own TPU custom AI chips with Broadcom (a 15-year relationship now exploding) and is spending $180B of capex this year — part of the "table stakes have exploded" hyperscaler arms race he calls existential but rational.
41:30— Okay. — Now we talk about AI um so they do two things. They do networking and they do custom chips — right — like Google for example makes what they call TPUs. This is a tensor processing unit. It's Google's own internal custom AI chips. Broadcom effectively works with them to make that chip.
In short: The AI business has become capital-intensive with "no moats," and spending ~$180B (vs ~$90B last year) forced an $80B capital raise — but there are "a lot of other reasons to own Google other than AI." Berkshire took $10B of the raise.
Eisman uses Google as his prime example that AI is turning even cash-rich tech giants "asset-intensive." Google spent ~$90B last year (coverable from its own cash) but is now spending ~$180B — so much that it had to raise $80B from outside investors, which it rarely needs to do.
He's not condemning the stock — "there are a lot of other reasons to own Google" beyond AI, and Berkshire just bought $10B of the raise — but he insists the AI business itself has "no moats" and now eats huge amounts of capital, which is the opposite of the easy-money software model investors got used to.
3:34I MEAN, I REALLY DON'T KNOW. I MEAN, LOOK, GOOGLE, THERE'S A LOT OF OTHER REASONS TO OWN GOOGLE OTHER THAN AI, BUT THERE'S NO QUESTION IN MY MIND THAT THAT THE BUSINESS HAS BECOME CAPITAL INTENSIVE, NO MOATS. AND THE OTHER THING THAT'S HAPPENING BEHIND THE SCENES IS THAT UNTIL NOW, COMPANIES HAVE BEEN CHARGING FOR TOKENS WAY BELOW THE COST OF TOKENS.
In short: Referenced as the AI-buildout backdrop: one of Broadcom's six core custom-XPU customers, and its ~$85B equity raise is part of why Broadcom is seen as a beneficiary. But Hock Tan also acknowledged Google will diversify its TPU supply — a key reason for the AVGO sell-off.
In short: Its 2004 IPO (~$23B) shows how IPOs used to start small and compound — unlike today's giant IPO valuations.
6:09But, of course, when so here's a good example. When Google went public back in 2004, I believe it was right around there, its total valuation, you know, including all the locked up shares was 23 billion. And if you went back and looked at Tesla, probably somewhere in that area. So my point is those things were starting with much much lower valuations.
In short: A warning signal, not a call: Google raised ~$80B via an equity private placement rather than debt — evidence the debt markets have made AI-capex financing too expensive (private credit "effectively frozen").
Dowd isn't rating Google's stock — he's using it as a warning light. Google raised about $80 billion by selling new shares ("equity") instead of borrowing the money ("debt").
Why does that matter? Borrowing is usually the cheaper way to fund big projects. Choosing to sell shares instead suggests lenders are demanding too much — a sign the debt markets are getting nervous about AI spending, and that the riskier corners of lending (private credit) are seizing up.
4:39So the AI bubble we believe is going to burst. The other thing that's interesting is I don't know if you saw, but Google issued 80 billion in equity uh uh private placement last night rather than using debt. So we think the debt markets are are are telling uh the AI companies that it's too expensive to use debt.
In short: Hyperscaler in the group whose combined CapEx fell quarter-on-quarter, with the year-on-year growth rate also easing in Q1.
Alphabet (Google's parent) is another of the five cloud giants in Woo's spending tally. Their combined AI-infrastructure spending fell versus the prior quarter and grew more slowly year-over-year — his evidence that the AI buildout's momentum is fading, not a call on the stock itself.
6:52The AI trade has always been driven by AI-related capital expenditures by the hyperscalers. Exactly what they're buying more of at any given moment may change, but the underlying driver remains the same. To me, the most important takeaway from the first quarter earning season is the fact that the combined CapEx spending of Microsoft, Amazon, Alphabet, Meta, and Oracle fell compared to the fourth quarter.
In short: Front-ran the SpaceX/Anthropic/OpenAI IPOs with an ~$80B raise to fund AI capex — to Larry a symptom of a force-fed, top-heavy market. (Danny reads the raise as bullish; Larry as a warning.)
Alphabet is Google's parent. It raised about $80 billion in debt to fund its AI spending, just ahead of the big SpaceX/Anthropic/OpenAI stock listings. To Larry that's a symptom of a top-heavy, "force-fed" market straining to absorb too much new tech supply. (The host, Danny, reads the same raise as bullish — that Google sees opportunity; Larry treats it as a warning.)
7:27So, we have these three big IPOs, right, which is about potentially $300 billion of of new stock coming on market. And guess what? Google front ran these guys today, you know, by putting in the 80 billion, right? So that's 80 billion of of cap they that they they need that for kind of like their AI capex and then the three IPOs like the divergences between say healthc care and and the market or healthcare and tech are way way getting stretched and so that's where we're starting to to really find some interesting value. Let
In short: Cited (with Amazon) as a swing factor for copper demand — if forecast data-center capex lands, the world would need more copper to 2050 than ever mined; he won't predict whether it happens.
32:38uh if the uh forecast investments in data centers put forward by you know people like Google uh and Amazon take place and I'm not saying they will by the way I'm not smart enough to know that but if they do uh we will need to produce more copper between now and 2050 than has been mined in recorded history and given the fact that we have underinvested in copper exploration for 30 years.
In short: The "Musk premium" foil — Google's quantum effort "barely registers in GOOG," whereas SpaceX gets credit for moonshots before they exist ("capitalized for one CEO and expensed for everyone else"). A disclosed author holding (GOOG).
In short: Straddles phase 1 (custom TPUs, why it's up now) and phase 3 (Search, YouTube, Gemini, Cloud, Ads). A core "buyer" bet — $130k in the passive-income portfolio plus another $90k in the Story Fund; he holds every share.
Alphabet (Google's parent) is unusual because it sits in two of Carlson's phases at once. It's partly "phase 1" — its in-house TPU chips are a scarce AI ingredient and a big reason the stock is up now — and partly "phase 3," the buyer/owner-of-the-customer phase where Google monetizes AI across Search, YouTube, Gemini, Cloud and ads for decades.
That dual role makes it one of his core bets: $130,000 in his passive-income portfolio and another $90,000 in his Story Fund, both well in the green, and he holds every share. His thesis is that once the AI sellers lose their pricing power, money rotates back to platforms like Google that "own the lifetime relationship" with billions of users.
12:32Now, when I look at Google, Google's one of these rare cases where they're kind of in phase one and they're kind of in phase three. Part of the reason that Google stock is so high is because of their custom TPUs. That's the reason it's going up today. But I believe that Google will go up even more as we get into phase three.
In short: Named (via Google Cloud) as a hyperscale buyer in NVIDIA's Hyperscale segment. The author owns GOOG in the App Economy Portfolio.
In short: A Fervo power-purchase counterparty — Google's geothermal offtake is cited as evidence that hyperscalers are buying firm geothermal power for data-center demand.
In short: Made "very conservative design choices" with TPU V8 and won't submit TPUs to its own MLPerf benchmark — but "I would never bet against Google," and TPU V9 "is going to be amazing" with more aggressive choices. Mixed → Neutral.
Google designs its own AI chip, the TPU, as an alternative to Nvidia. Baker's view is genuinely mixed. On the cautious side: he thinks Google made "conservative" design choices with the current TPU (V8), and pointedly notes Google won't submit its TPUs to MLPerf — the very chip-benchmark Google itself created — which he reads as a tell.
On the positive side: "I would never bet against Google," and he expects the next chip, TPU V9, to be "amazing" with more aggressive design. Net-net it's a balanced, wait-and-see stance rather than a clear buy or sell — hence Neutral.
17:15And if somebody is wildly bullish on TPUs today, let's go look at their 13F and let's see if they owned Momentum or Elastica, which were the best ways to invest in TPUs. I owned one of those. And so I do feel like I have some credibility to say this, but I do think Google for a lot of reasons made very conservative design choices with the TPU V8.
In short: The poster child of the Other Income circularity: booked $38bn of Other Income in 1Q (with Amazon, $53bn combined ≈ 60% of their net income; $49bn from private-stake revaluations). Alphabet is the largest investor in Anthropic — whose $200bn compute commitment to Google both grows Google's revenue backlog and, via successive up-rounds, marks Google's stake higher into "earnings." Cited as evidence, not a fresh stance (his standing GOOG-long view is separate).
Here's the trick Paulo is pointing at. Google puts money into a private AI company (Anthropic). That company's "value" keeps getting stamped higher every few months in new fundraising rounds. Because Google owns a piece, it gets to write up the value of its stake and report that paper gain as profit — "Other Income" — a whopping $38 billion of it in one quarter. At the same time, Anthropic promises to spend $200 billion buying Google's cloud services, which boosts Google's sales. So Google's own investment inflates both its revenue and its reported earnings, even though little real cash changed hands. Paulo isn't turning bearish on Google specifically; he's flagging that a big chunk of Big Tech's celebrated "earnings" is really a mark-up on stakes the companies themselves are pricing.
In short: A "follow the cash" example — search built huge value with ~zero capex; now its legacy cash is being recycled into data-center infrastructure, redirecting trillions toward commodity/infra suppliers.
Google is his "follow the cash" example. Search built enormous value while spending almost nothing on physical equipment. Now that mountain of legacy cash is being plowed into data centers and infrastructure — redirecting trillions toward the companies that supply commodities and build things. The takeaway: the suppliers, not the tech platforms, are the next 20 years' winners.
13:29While we were doing this and having this explosive growth in these areas of the economy, we were going through this deindustrialization period where we were literally tearing down our industrial footprint to move it over to China. And effectively just destroyed the domestic supply chain in the US. And now the world has done a 180° pivot because if you look at these same tech companies that created so much value, today the business model is whatever cash our legacy business generates.
In short: Listed among Arista's customers ("Microsoft, Meta, Google, Oracle…") and among the hyperscalers driving the end-market tailwind. No view here — Alphabet had been added to the rated Buy list a week earlier, on 7 May.
In short: His big-name pick — "they've won AI… the only public company with a foundational model." TPU chips "phenomenal" (now powering Anthropic), search accelerating, YouTube/Gmail/Sheets to infuse AI; "the stock is very cheap… could easily be up 50%. I don't see very much downside."
Google is his big, easy-to-own pick. He says it has "won AI" and is the only publicly traded company with its own top-tier foundational model (Gemini). Its in-house AI chips — TPUs — are "phenomenal" and now even power rival Anthropic, which is a strong vote of confidence.
On top of that, AI is speeding up search rather than killing it, and Google can fold AI into YouTube, Gmail, and Sheets. With the stock "very cheap" and revenue set to accelerate, he thinks it "could easily be up 50%" with little downside.
21:55Then they have 40% of their business which is defense. And there's a huge up cycle in defense. They've won business with the Iron Dome contract, and defense is getting so electronicized. The second one is just Google. It's simple. They've won AI. They're the only public company with a foundational model.
In short: NEWLY ADDED to the Buy-Hold-Sell List — "global technology holding company", and one of the three big-tech April performers cited (+21.7%, "strong growth in AI and the core advertising business"). Rated BUY: EPS growth 15.0%, FWD PE 29.9 against a fair exit 25.0, expected return 13.6%, fair value 665.0 against a 381.8 price = 42.6% undervalued. Note the tension with 21 April, where the stated entry was 18x forward — $210 against a $336 price.
Alphabet is added to the watchlist this month and rated a Buy: the model puts fair value at $665 against a market price of $382, an expected return of 13.6% a year, on assumed profit growth of 15%.
It is also one of the three big technology companies whose April surge is used, earlier in the same issue, to illustrate how concentrated and AI-driven the market has become — up 21.7% in the month on AI and advertising strength.
There is a tension a careful reader should notice. Sixteen days earlier the stated willingness to buy Alphabet was at 18 times earnings, or about $210 a share, against a market price of $336 — a refusal to pay up. The model here reaches a Buy at $382 by assuming the shares should eventually trade at 25 times rather than 18. The company has not changed; the assumed exit multiple has.
In short: "Just a really powerful quarter" — total revenue 94.7B vs 91.6B est; cloud 20B vs 18B est, a meaningful acceleration. Stock +10% Thursday. "That's all you really need to know."
On this week's earnings alone, Google delivered "just a really powerful quarter" — revenue of $94.7B beat expectations, and crucially its cloud-computing business accelerated to $20B versus $18B expected. Faster cloud growth is the clearest sign that corporate AI demand is real and rising. The stock jumped 10%. "That's all you really need to know."
21:49Google's total revenue was 94.7 billion. I'm going to say that again. For the quarter, Google's total revenue was 94.7 billion versus 91.6 billion expected. Google's cloud computing revenue of 20 billion versus 18 billion expected, so a meaningful acceleration in growth. And that's all you really need to know. Microsoft, revenue was 54.5 billion versus 53.8 billion expected. That's good. Earnings per share was 427 versus 405 expected and versus 346 last year, which equates to a very impressive 23% growth, also good. And most importantly, its cloud computing business, Azure, saw 39% revenue growth versus 38% expected. And the company gave strong guidance on the call.
In short: The widest gap on the list. "Alphabet is an amazing company, but it also did really well recently. The stock is up +123% in the past year. As a result, the valuation looks rather expensive" at a 28.9x forward PE. "I would love to buy Alphabet at a Fwd PE of 18x. This would mean at a stock price of $210 (current stock price: $336.0)" — a 38% fall required. Case: a search-advertising duopoly with Meta, network effects and data scale, a self-reinforcing user/data/advertiser loop, and diversification into Cloud, YouTube and other bets, with advertising still ~75% of revenue.
Alphabet is Google: search advertising and YouTube, plus a cloud computing business and the Play Store. Advertising is still about three-quarters of what it earns, and it shares that market with Meta in what is effectively a two-firm arrangement.
Nothing about the business is in question here. More users produce more data, better data makes the ads work better, better ads bring more advertisers and more money for search — a loop that is very hard to break into.
The objection is purely the price. The shares have risen 123% in a year and now cost 29 times next year's profits. The stated willingness to buy is at 18 times, which works out at $210 against a market price of $336 — a fall of nearly 40%. That is not a prediction; it is a refusal to pay up, written down in advance so it can be checked later.
In short: The most widely held name on the page after Brookfield: Markel 6.9% and Brederode's largest listed position at 15.1%. Structural reference; no stance is offered here.
In short: The other communication-sector hyperscaler inflating S&P concentration; cash cow → capital-intensive.
Alphabet (Google) is the other hyperscaler he names in the same breath as Meta.
Same story: a former cash cow now spending heavily on AI infrastructure, and another big weight inflating the S&P's tech concentration. He'd diversify away from this crowded top end of the index.
26:34Now, the composition in technology is really not 35%. It's close to 50% of the index in artificial intelligence technology if you count the communication sector, right? Cuz that communication sector includes Meta and Google. And remember, these hyperscalers — Meta, Google, Microsoft — these are companies that have been cash cows for the last 30 years. And now all of a sudden they're very capital intensive businesses.
In short: On the quality-fund trim list on valuation ("did really well recently"), while ranking #9 on the two-quarter aggregate Top Buys table. Both facts reported without a house view.
In short: Named as the incumbent in Tesla's most-hyped end market. "Waymo (Google): The leader in Robotaxis, with driverless cars already picking up passengers in cities like Phoenix and San Francisco." The framing matters — the robotaxi option embedded in Tesla's 201.2x multiple is up against a competitor already carrying paying passengers. No stance on Alphabet itself here; the standalone case is made on 21 April.
Waymo is Alphabet's self-driving subsidiary, and it is named here as "the leader in Robotaxis, with driverless cars already picking up passengers in cities like Phoenix and San Francisco."
That single fact bears directly on the Tesla valuation. A large part of the 201-times multiple rests on robotaxis becoming a real business for Tesla in the future — while a rival is already running the service commercially today. No stance is taken on Alphabet in this issue; the standalone case for it is made in the April shopping-list series.
In short: Bill/Cole — the concentration & hyperscaler-capex critique. The biggest holding in the Russell 1000 Value index ("even value doesn't know what it is"). One of the hyperscalers going from asset-light to capital-intensive — returns on capital declining even as revenue grows.
16:14No, it they because it trades at a discount to the S&P. It's the most expensive S&P in in my 45 years in the business. They they think, "Oh, this company is is 20 times earnings, therefore its value." — Let's talk about two areas of the market. Um, and I want to get I think I know what you're going to say, but let's talk tech.
In short: The verb-as-moat analogy — "to Google" (like "to Xerox") became a verb, and both were "massive winners in their heyday"; Google now "continues to make hay in a big way" at a ~$4T market cap. Cited as the precedent for Uber's noun-becomes-verb dominance, not a fresh call on Alphabet.
In short: #7 pick. Carried entirely on other people's conviction: Munger in 2009 — "Google has a huge new moat. In fact, I've probably never seen such a wide moat" — and the follow-through, "Buffett and Munger said multiple times that they regret not buying Google earlier. Well… they finally did recently. Berkshire Hathaway bought 18 million Google shares in 2025." Table: 32.2% net margin, 29.5% ROIC, 30.2x forward, 15.0% expected growth. No Compounding Quality position or valuation view.
In short: #6 most-picked; +65.2% — the year's clearest sentiment reversal, reported without a house view. "Google's search faces its biggest threat: people are turning to AI chatbots like ChatGPT… fewer searches mean less ad money. But today, many believe Google is leading the AI race. OpenAI CEO Sam Altman has declared a 'code red' as the company scrambles to catch up with Google's Gemini 3 models."
In short: Referenced as the benchmark he holds and his prior high-conviction "asymmetric bet" — he made "those type of bets with Google in early 2025" and is now doing the same with Mastercard, which is "my biggest position, even bigger than Google." Also a growth-rate yardstick — VAS growing "faster than Google."
2:40That's a lot of money. That's my biggest position, even bigger than Google. Now, it's not my biggest winner, but it's a company that I've recently bought a lot in because I feel so confident about the company. Now, the way that I would just underwrite this investment very quickly is that Mastercard is not going to be the best performing stock of 2026.
In short: The historical asset-light winner — Brin & Page "didn't have to do the capex to build the internet," so Google won in 2002. Cole's point cuts the other way for today: "you don't want to own the people that do capex," and the hyperscalers (Google included) are now becoming capital-intensive. A cautionary reference, not a buy.
Smead uses Google as his key historical example of the right way to win a technology boom: back in 2002, Google succeeded precisely because it didn't have to spend money building the internet — it just ran its software on top of what others had built. His lesson for today is the opposite of a buy signal: "you don't want to own the people that do capex," and the big tech companies (Google included) are now shifting from asset-light to spending enormous sums on AI infrastructure — which historically drags returns down. So he cites Google as a cautionary reference, not a pick.
3:11— THAT THEY DIDN'T BUILD THE INFRASTRUCTURE. GOOGLE ANDREW. SO SO THINK ABOUT IT LIKE THIS I WAS WATCHING ERIC SCHMIDT SPEAK HERE IN PHOENIX LAST WEEK. AND THINK OF YOUR IF YOU'RE IF YOU'RE SERGEY BRIN AND LRY PAGE, YOU'RE SITTING THERE AT STANFORD IN OH TWO AND YOU'RE HAVING A CUP OF COFFEE AND YOU'RE LIKE, ISN'T THIS INCREDIBLE? WE CAN RUN OUR ALGORITHM ON DIGITAL SEARCH IN A WAY THAT'S NOVEL FOR ITS TIME.
In short: The winner of the turn: Gemini 3.0 "caught up and passed OpenAI" without NVDA GPUs or a Stargate/Oracle financing morass; ~$70bn FCF/4 quarters, institutional memory, no key-man risk. Berkshire bought GOOG in Q3 at ATHs. Framed as the long leg of a GOOG-long / NVDA-funding-short pair.
For two years the AI story was "OpenAI and NVIDIA win everything." Paulo argues that just flipped: Google's new Gemini 3 model caught up to and passed OpenAI — and Google did it with its own chips and its own cash, without the expensive borrowed-money data-center schemes the others rely on. Google throws off about $70 billion of spare cash every few months, has deep engineering culture, and doesn't depend on one founder.
So he frames it as a pair trade: own Google (the durable winner, which even Warren Buffett's Berkshire bought at record highs) and short the names that need constant new funding to keep the story alive. Google is the "long" side.
Nothing matches this filter.
Verbatim excerpts from the public transcripts (auto-pulled at each mention's timestamp, lightly cleaned). Timestamps link into the video; "source page" opens that commentator's full analysis at this row. Click a mention's header line to expand it (one open at a time). For personal study — not investment advice.