In short: Link owns it and points to the de-rating (20:12–20:28). AWS growing 37% (Google Cloud 82%, Azure 43%) — monetization is showing up — yet "I own it, of course. It's at 20 times, it's 10 year historical average's 44 times."
Link owns Amazon and points out that its cloud business (AWS) is growing 37%, so AI spending is starting to pay off. Even so, the stock trades at about 20 times earnings, less than half its 10-year average of 44. To her, that is a strong business at a discount.
In short: Referenced only — the counterparty in GNRC's up-to-$8B backup-generator supply agreement ("the largest e-commerce and cloud infrastructure company in the world"), holding the right to buy up to 1.7M GNRC shares at $200.93 as it places orders. Cited as the validation of the GNRC thesis. No view on Amazon shares.
In short: Blocked Muse ("unauthorized AI agent"). He calls the credentials excuse "entirely bogus," since a logged-out Muse is blocked too. The real motives: agents make the sponsored listings behind its $76B high-margin ad business "less valuable," they disintermediate its customer touch point, and Muse bypasses its own Rufus shopping assistant. "Amazon is clearly exposed with the ad business." He expects the block to be temporary, the start of a negotiation: a revenue-sharing deal "is the best deal for both Amazon shareholders and Meta shareholders," even "at the jeopardy of some ad revenue."
When you search on Amazon, the first results are often "sponsored": sellers pay Amazon to put their product at the top. That advertising brought Amazon about $76 billion over the past year, and it is very profitable. It works because humans look at the top of the page. A shopping bot like Muse doesn't care what's on top. It checks every listing and picks the best deal, so those paid spots lose their value.
That, not password security, is why Carlson thinks Amazon blocked Muse. Amazon also loses its direct relationship with the shopper (what you click, what you like) and watches Muse bypass its own shopping helper, Rufus. He expects the block to be a bargaining move. Both companies lose if it lasts, and Walmart could welcome the bots instead, so he expects some paid arrangement eventually, even if it costs Amazon some ad money.
13:11It's browsing Amazon as an anonymous person, not logged into anything. So, if security and credentials and safety and privacy are not really the concerns that Amazon's worried about, then what are? Well, Amazon actually has good reason to be concerned about Muse that go far beyond security or saving your password. Amazon made $76 billion in advertising over the past 12 months. 76 billion.
In short: Passing mention — named only as an expected data-center counterparty on LandBridge's Texas acreage; no view on Amazon.
33:55So we have a position in it just as a disclosure and it's a company that everyone thinks that they do only oil but actually they just rent the right to drill and use the oil, it's actually the water, and that is in the field. So they would, and right now there would be a lot of data centers in Texas and you expect them to sign a contract with Amazon and then it's a Meta and everything and it adds to their earnings that are really really good and they are like 30 people that work there.
In short: Named only as one of the well-capitalized buyers that will "gobble up" permits, supply and contractor capacity from spec data-center builders (RPK: "do you have a permit? All right, we'll just take that project off your hands"); Sam also raises investment committees comparing Amazon/Google credit with sovereign credit.
55:30And so anyone who has permitting — even if you can't get [power] — you'll be in a position where they're buying up permits. Amazon or somebody will step in and just buy — do you have a permit? All right, we'll just take that project off your hands, because we'll do it down the road. We're not in a hurry.
In short: Referenced only — as a sector-classification puzzle: filed under consumer discretionary, but the retail side is "almost consumer staples" (he bought Ortho ant killer there) and AWS "is truly the future of the business".
25:00Now, discretionary is its own weird animal in that we'll see how things shake out in the coming years, is this company discretionary or is it tech, but generally you get Amazon and Tesla inside consumer discretionary. Well, I mean, what do you think, Jeremy? I kind of think Amazon's, well, it's Amazon Web Services plus the retail business, but the retail business is almost consumer staples.
In short: The other name Baruch is buying more of (25:59) — "we're buying more Amazon… to lean into that right now," to own the drivers of the Mag 7 ETF's breakout to a new high.
In short: Mixed: product search and sponsored listings are exposed — Muse "will shop on Amazon plus a million other websites." But selection, Prime, fast fulfillment, returns and AWS mean "agents may prefer Amazon because the full transaction is reliable," and AWS can run the agents; it is building its own shopper in the new Alexa. Risk to search/ad economics, opportunity in more transactions.
Amazon has been the default place to start shopping. An agent that searches every website at once weakens that, and could hurt the paid product listings Amazon earns from. But Amazon also has the huge selection, Prime, fast delivery and easy returns an agent would want for a reliable purchase — plus AWS, which can run the agents themselves. Some of the business is hurt, some helped.
30:03That weakens Amazon's role as a default place to begin a product search and may pressure sponsored listings if the agent filters aggressively. Yet, Amazon owns enormous seller selection, prime loyalty, fast fulfillment, returns, payments, reviews, advertising demand, and Amazon Web Services. Agents may prefer Amazon because the full transaction is reliable.
In short: Neutral on the equity, positive on the credit: a hyperscaler whose AI borrowing is price-insensitive ("6, 8, 10, 12 — it doesn't matter") yet whose bonds are "probably money good." Only one or two of the five will win the build-out race, which is an equity risk, not a bond risk.
Amazon is one of the "price-insensitive" borrowers: it will keep borrowing to build AI capacity whether rates are 6% or 12%, because it sees the race as existential. Bassman thinks that pushes up interest rates for everyone, the US Treasury included. He still thinks Amazon's lenders get paid back from its profitable core businesses. The stock carries the risk that the spending doesn't pay off.
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: Passing mention: signed "behind the meter" nuclear power deals, is a financial backer of X-Energy, and (news blurb) is buying data-center generators from Generac in a $2.4bn deal.
In short: AWS acceleration as ROI evidence (Link). Link's third cloud data point: "Amazon AWS… accelerate last quarter substantially." Her sizing of the build the ROI has to cover: "$7.1 trillion on data centers between now and 2031, $1.4 trillion on the grid… power, 1.7 trillion… we just don't have enough of it."
In short: Split between time frames. The negative free cash flow "bothers me a lot," and near term "if oil goes to 120 bucks, Amazon has a consumer business… Amazon will struggle," with oil high "at least through the midterms." Longer term "Amazon's very interesting": "the next big wave… is robotics and Amazon is ridiculously good in that," and AI plus robots make the asset-heavy logistics core "a lot more efficient."
Niles is torn on Amazon depending on the time frame. Longer term he finds it "very interesting" because robotics — which he calls the next big wave after AI — should make Amazon's huge network of warehouses and delivery far cheaper to run, and Amazon is "ridiculously good" at it.
Shorter term, two things bother him. Its free cash flow has turned negative because of AI data-center spending. And much of its business is selling goods to consumers: if oil goes to $120 a barrel — he expects high oil at least through the November midterms because of Iran — shoppers get squeezed and "Amazon will struggle."
52:22For the first time in years going back to burning cash as, I think they're one of the biggest spenders on AI from a capex perspective. — Yeah, don't get me wrong, that bothers me a lot. Now, here's the one thing with Amazon, though. I think the next big wave, which is kind of tied to AI obviously, is robotics and Amazon is ridiculously good in that and Amazon has this asset-heavy model as well.
In short: Risk isn't just position count: Amazon is "technically one company, but you could almost treat it a little bit like an ETF" given its many cash-generative businesses. Also among the names he's "excited about."
In short: The third Lebenthal hyperscaler — the one that didn't hold up. "Now, Amazon's obviously red, we see that." Held as part of the spender basket Santoli calls the port in the storm; no separate case made on the day.
In short: Named with Microsoft as the debt issuers funding AI capex — part of $3T+ of off-balance-sheet commitments "the bond market is rebelling against."
Amazon comes up alongside Microsoft as a big AI spender that raises money by selling bonds. Together these companies have promised more than $3 trillion of future spending that doesn't show up on their balance sheets, and Woo says the bond market is "rebelling" against the debt coming. Again it's evidence for his rates view, not a call on Amazon itself.
21:51for Trump to taco, for oil price to stop going up. So if the stock market doesn't go down the oil price is going to keep going up until the stock market basically pops. The same thing with interest rates, right? Right now interest rate is trading on this whole AI capex story, right? If the stock market doesn't go down, that means Microsoft, Amazon, it's going to continue to issue debt to finance their AI capex and rates are just going to keep going higher until the stock market pops. So at this point, that
In short: "Amazon spends a lot of money on AI, but they have AWS that directly monetizes AI." The first of the three hyperscaler comparisons in the slowdown screen: high capex is not the tell — high capex without a direct revenue stream is.
4:40All of it is regulated. If that is the case, what company is spending more money on capex, more money on AI development with less direct revenue from AI than Meta? Right now, there really isn't one. Amazon spends a lot of money on AI, but they have AWS that directly monetizes AI. We have Google with Google Cloud.
In short: Same point as Google: a multi-trillion-dollar customer incentivized to design its own chips to avoid bottlenecks and the "Jensen tax."
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: Held per her Sep 4 portfolio post, as read out by the host — not discussed as a holding (Amazon appears only as the failed Roomba acquirer).
In short: The other side of the Qualcomm deal: receives warrants on 25M QCOM shares worth >$4B while partnering on custom AI chips — the hyperscaler building its own silicon to reduce NVIDIA dependence. A subscriber suggests "Amazon's contribution of IP" explains "Amazon's asking for warrants." No view on Amazon itself.
16:01There was a very interesting announcement this week from Qualcomm. Qualcomm announced a collaboration with Amazon to build custom AI chips and optical networking equipment. The deal is designed to create competition with Nvidia. As part of the deal, Qualcomm issued a warrant to Amazon giving Amazon the right to buy 25 million Qualcomm shares at $161.26.
In short: One of the data-center builders his audience owns; hacked "every single day" by state actors yet defended — AI helps defenders as well as attackers.
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: Named only inside the hyperscaler group — "the hyperscalers, Google, Microsoft, Amazon, and Meta, have spent over a trillion dollars in capital expenditure buying GPUs from NVIDIA" — and so carries the group's exposure: "OpenAI and Anthropic are estimated to make up around 70% of hyperscaler AI revenue." No AWS-specific figure is given in the captured section, and no view on the business itself.
In short: The buyer in the day's deal, discussed as a hyperscaler rather than as a stock. Belski frames the AWS tie-up as the validating counterparty for Qualcomm ("a very good signal that they're getting in business with AWS"), while Terranova reads Amazon's side as routine customer-locking: "this is not a unique deal you're seeing. A lot of the hyperscalers like Amazon going out trying to get relationships to secure these customers." No view is expressed on Amazon shares.
In short: Cited twice, both times as an example rather than a call. Dawson pairs it with Alphabet as a GAAP-only reporter whose headline earnings carry unrepeatable one-time securities gains into the 2027 comps. Forehand uses it as the canonical survivorship-bias case against buy-and-hold: "there's a bunch of survivorship bias in this idea that you just needed to buy and hold Amazon… let's look at the other companies that looked like you needed to buy and hold them at the beginning before Amazon looked like that, and you did not need to buy and hold those."
Amazon shows up twice, in two unrelated arguments, and in neither is it a recommendation.
In Dawson's accounting segment it is grouped with Alphabet as a GAAP-only reporter — the same one-time-gains distortion, the same 2027 comparison problem described above.
In the buy-and-hold segment it is the poster child for survivorship bias: everyone points to Amazon as proof you should buy great companies and never sell, but the reason that story is told is precisely that Amazon survived. At the time, it looked like the dozens of contemporaries that did not. The lesson offered is about the sampling, not about Amazon.
29:41I think you want to have a very flexible open mind and as the facts change, you want to change too. — We don't get a lot of "moronic" here. — His quote was saying there's some stocks you just need to buy and hold is completely — which by the way is true. There's a bunch of survivorship bias in this idea that you just needed to buy and hold Amazon.
In short: A structural reference inside the Meta argument, not a view. Nathan names it first as the monetization advantage Meta lacks — "they don't have that big cloud business like Amazon, like Microsoft, and like Google" — and then as the venue where the commoditization lands: the models "are all going to be sitting on AWS or Azure or whatever other cloud platform, and you're going to be able to choose anyone you want… you're going to choose the cheapest one from a token perspective." The implication carried but not stated: the cloud layer, not the model layer, is where the toll is collected.
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: 12.6% / $187k (+$62k). 18% EPS growth × 30× = 13.48% — lowered because the stock is +13.5% YTD. AWS adding 2M more Nvidia GPUs in 2027–28 as "demand has exceeded its earlier expectations," Prime Air drones to ~500 US towns (6×), Alexa+ in tens of millions of homes. "One of the strongest stories." $1,000 (10%).
Amazon has one of "the strongest stories" to him: its cloud unit is adding millions more Nvidia chips because demand beat expectations, drone delivery is expanding six-fold to about 500 US towns, and the Alexa+ AI assistant is spreading. Because the stock is already up this year he trimmed his expected return to about 13.5% a year — still strong for a company this large.
In short: Named twice, from opposite directions, by people who do not discuss the stock itself. As evidence for the bull case, Lebenthal quotes the CEO: "Andy Jassy, CEO of Amazon, saying things like it takes less than three years to recoup the cost of a new data center" — one of his three proofs that the AI CapEx is profitable. As a competitive risk, Baruch's Walmart pitch argues the retailer is "eating into Amazon's core business largely unnoticed." No committee stance was taken on the shares.
Amazon appears twice without anybody taking a position on the shares, and the two mentions point opposite ways.
As evidence, its chief executive's claim that a new data centre pays back in under three years is one of Lebenthal's three proofs that AI spending is profitable rather than speculative — a payback period is the plainest possible test of that.
As a risk, Baruch's Walmart case rests partly on the argument that Walmart is "eating into Amazon's core business largely unnoticed." Both can be true: the cloud business compounding while the retail business faces a sharper competitor.
In short: Context only: AWS is one of "the biggest cloud platforms in the world" whose combined growth went from 35% in the March quarter to 43% in June, with "operating margins expanding by 2%." No stock view.
21:42And if you look at the biggest cloud platforms in the world, so Amazon Web Services, Microsoft's Azure and Google Cloud Platform, if you look at the growth rates for the cloud businesses that they have, they were growing 35% in the March quarter and that accelerated to 43% year-over-year growth in the June quarter.
In short: AWS is named first among the "big three public cloud vendors" whose growth accelerated to 43% with operating margins up ~2 points — the profitability half of his bubble test. Amazon is also his base-rate exhibit for winner-take-most ("who's the market share leader in e-commerce? Well, it's really only Amazon") and one of the companies built on the wreckage of the last buildout's cheap bandwidth.
AWS is Amazon's business of renting out computers, storage and AI capacity. It is the first name on his list of the "big three public cloud vendors," and it is part of the evidence that the AI build is currently profitable rather than merely large: growth accelerated and margins widened in the same quarter.
Why both matter together: a cloud provider can always buy faster growth by cutting prices or eating the cost of the chips, and that shows up in the margin. When growth speeds up and the margin widens, customers are paying more for capacity that is genuinely scarce.
Amazon also does duty as his historical base rate. When he argues that technology markets end up with one dominant winner rather than ten healthy ones, e-commerce is his cleanest example — "it's really only Amazon, right? Then you have a bunch of smaller players." And it is one of the businesses that only existed because the previous bubble left behind cheap infrastructure, which is the shape he expects cheap AI tokens to produce next.
4:57The second part of that is how profitable is this? Because you may be generating revenues, but if you're losing tons of money, it really doesn't matter. The good news is profitability is also improving. And if you look at the big three public cloud vendors, and so that would be obviously Amazon's Web Services, Microsoft's Azure, and Google Cloud Platforms.
In short: "Amazon… is probably up something like 10x since 2016 or 2017" — you made extreme returns owning the most well-known company in the world, and "that profile of return… is probably behind us." AWS reaccelerated from a hoped-for 18% to the 40s and the stock barely moved, which he reads as queasiness about ROI and customer concentration. Still 15–20%/yr possible; just a different profile.
Amazon is Keller's illustration of how the megacap trade has changed. The stock is up roughly ten times since 2016–17 — you got extraordinary returns for owning one of the most famous companies on earth. He thinks 15–20% a year is still achievable from here, but that the multiple-bagger profile people mentally attach to these names is behind us.
The telling detail is AWS, Amazon's cloud business. Two years ago investors were hoping it could grow 18%; it's now growing in the 40s — and the stock is only incrementally higher. That gap between spectacular growth and a muted share-price response is, to him, the market quietly expressing unease about how much of that revenue converts into durable profit, and about how few customers it now depends on.
28:02that it's not going to be the easy trade to outperformance. — I think so. I own some of them. The stocks, I think, are reasonably cheap. I think it's just a very different profile than it was over the number of years. Amazon, you can check me on this, but is probably up something like 10x since 2016 or 2017.
In short: Present three times, never as a recommendation. Demand: "AWS announced a major multi-year expansion to deploy 2 million additional Nvidia GPUs across its global infrastructure in 2027-2028." Earnings optics: Amazon is one of the two names distorting the index — "excluding Amazon… would reduce the consensus consumer discretionary EPS from a record 92% to just 7%," and "excluding Amazon and Alphabet from the S&P 500 would reduce the overall earnings surprise rate from 29% to just 10%," because of "how big Anthropic's mark-to-market was for Amazon and Google's P&L." Automation: "Amazon is developing fully automated delivery stations through Project Tetromino… focused on automating the final warehouse stop before packages reach drivers" (a 2028 pilot at ~2.5× current throughput, >$530M of investment by 2029, which Amazon disputes). And it is one of the seven inside the $3.1 trillion of off-balance-sheet arrangements.
Amazon appears three times without being a recommendation. Its cloud division committed to installing two million more Nvidia chips across its data centres in 2027 and 2028 — one of the largest single orders behind Nvidia's forecast.
Second, and more awkwardly, Amazon is one of the two companies distorting the entire American earnings season. It owns a stake in Anthropic, whose valuation has risen enormously, and accounting rules require that gain to be booked as profit even though no shares have been sold and no cash received. The effect is startling: excluding Amazon alone would cut expected earnings growth for the whole consumer discretionary sector from a record 92% to 7%. Excluding Amazon and Alphabet together drops the S&P's earnings-beat rate from 29% to 10%.
Third, it is automating: Project Tetromino aims at fully automated delivery stations — the last warehouse stop before a package reaches a driver — targeting roughly two-and-a-half times the throughput of today's sites, with a 2028 pilot. Amazon disputes the reported figures.
None of this is a view on the shares. It is Singh assembling the evidence for two arguments: that headline index earnings are flattered by paper gains that could reverse, and that the AI build-out is real enough to be measured in physical orders.
Full passage: premium transcript (PDF).
In short: Referenced as part of the hyperscaler class, not named individually here. Carries the same 25–35%-of-cloud-revenue exposure to two loss-making private counterparties that defines the group's risk in this conversation.
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; "positioned to be more durable" than Nvidia. Nvidia's report says AWS will add another 2M Nvidia GPUs.
In short: Buy target $200 ("one of my favorite companies ever"; a $191k / 12.8% position, +$67k). Third of a near-evenly-weighted top three — Google 13.8%, Mastercard 13.3%, Amazon 12.8% — where "whatever one has a good week or a good month… will leapfrog to the top place position." He allows Amazon the fastest assumed growth of the megacaps at 16.1% EPS ("they have a tremendous amount of operating leverage… more of a growth stage than the other companies") on a 27 multiple: at $200 that's a 19.3% CAGR, "that money would around 2.4x over the next 5 years."
8:54So far, Google has held its own for quite some time, but Amazon has been working its way up more recently. Amazon is a $191,000 position, $67,000 of that being gains. Now, when I look at Amazon, the buy-in target for this one is at $200 per share. If Amazon reaches $200 first, then this company's getting the $10,000. And when we look at the assumptions that I believe are somewhat conservative.
In short: Another of the five capex-funding issuers and Nvidia customers, with two sell-side datapoints attached: "Morgan Stanley sees Amazon with a bull case path to 500 by 2027 driven by AWS potentially scaling towards 1 trillion in annual revenue over the next eight to 10 years," and a commitment to "invest more than two billion across Latin America through Prime Video, focused on original content and live sports."
Amazon is another of the five hyperscalers absorbing Nvidia's price increase and funding data centres with debt. Two other items attached to it this week: Morgan Stanley sketched a path to $500 a share by 2027 on the possibility that AWS, its cloud division, eventually reaches $1 trillion of annual revenue over eight to ten years; and Amazon committed more than $2 billion to Latin American content and live sports for Prime Video.
Neither is presented as a reason to buy. The cloud number is the bull case that justifies the spending, and the content spending is a reminder that the capital budget keeps growing on several fronts at once.
Full passage: premium transcript (PDF).
In short: Tied with TSM as the most widely held stock — a top-five holding at 9 of the 20 funds — and one of the three "mega-cap platforms" (AMZN, GOOG, META) in the ten names representing nearly half of all top holdings. The single most telling move: Druckenmiller "almost reversed last quarter's mega-cap retreat," rebuilding Amazon to 542,000 shares after nearly eliminating the position in Q1 — "showing just how quickly his positioning can change."
Amazon is two businesses in one: the retail marketplace everyone sees, and AWS, the cloud-computing arm that rents out servers and now AI capacity — where most of the profit comes from.
It is tied with TSMC as the most widely held name across the 20 funds, a top-five position at nine of them. The move worth studying is Druckenmiller's: he had nearly eliminated Amazon in Q1 and this quarter rebuilt it to 542,000 shares, alongside reopening Alphabet.
App Economy's own gloss on that reversal is the lesson rather than the trade — it shows "just how quickly his positioning can change." A quarterly filing is a photograph of a portfolio that may already look completely different by the time you read it.
In short: The proof-of-concept for the whole holding argument, cited twice: "David Gardner held Amazon when it was down 95%. Most people are not able to do that." Used as evidence, not as a view — Heyndrikx does not say he owns it.
In short: "I still like Amazon" — cloud doing relatively well and its own TPUs easing the GPU bottleneck, but it sits inside the cash-burning hyperscaler cohort whose doubled cloud backlogs lean on OpenAI and Anthropic.
Still a name he likes, with a cloud business doing well and its own in-house chips easing the GPU shortage. The caution is shared with the rest of the group: its cloud order book has doubled, but a meaningful chunk of those orders come from AI labs whose money was itself raised from investors — so the backlog is less independent than it looks.
18:54I still like Microsoft. I still like Amazon. I'm just saying that the businesses are not the capital-light businesses they were before. — Take a look at this article from the FT yesterday. Goldman Sachs analysts have scoured through the footnotes of the hyperscalers' regulatory filings, counted 1.
In short: Relayed as an Ackman trim, read as profit-taking rather than a view change: "He slightly reduced Amazon. This one has done really well, so I believe he's taking some gains out of this one and simply reallocating it." Carlson passes no judgement on it here.
17:04He added more to Restaurant Brands International. Finally, we get further up the holding list for Bill Ackman, and we get to companies like Amazon. He slightly reduced Amazon. This one has done really well, so I believe he's taking some gains out of this one and simply reallocating it. When we look at Microsoft, this is another one that he added to recently, another 10%.
In short: Both halves of the argument land on it. Trennert's client: "I've never heard a more bullish conference call in my life than Amazon last night" — "and it's true, but I don't see how you do it without hitting the bond and equity markets… it's not going to come out of cash flow" (Amazon's cash flow "was negative for the 12 months"). Verrone's tape: "Amazon just made new highs," part of the two-to-three-week momentum in the hyperscalers that makes them "at a minimum… tradeable." Also Trennert's example of the tech-adjacent name that pushes S&P tech-plus-tech-adjacent weight "over 50%."
Amazon appears three times, and the three appearances are the episode's argument in miniature. As a business, its cloud arm is one of the five hyperscalers building AI capacity. As an accounting object, its free cash flow — cash left after capital spending — was negative over the trailing twelve months, because the capital spending is enormous. As a stock, it just made new highs.
Trennert's client heard "the most bullish conference call in my life," and Trennert's reply is the honest version of the bull case: the demand is real, "but I don't see how you do it without hitting the bond and equity markets… it's not going to come out of cash flow." That is not a bearish call on the business. It is a statement about who funds the build — increasingly bondholders and new shareholders rather than the company's own earnings, which dilutes the existing owner.
Verrone's contribution is the reminder that the market has already absorbed a version of this: the name fell hard from mid-May and has since recovered to new highs. Net positive, with the funding question unresolved.
21:26He's like, "I've never heard a more bullish conference call in my life than Amazon last night." And it sounds like they're going to need — they're going to spend way more on capital and they're going to need to spend a lot more on capital spending, all the rest of it. And it's true, but I don't see how you do it without hitting the bond and equity markets, right? Right? It's not going to come out of cash flow.
In short: The other half of the mark-up problem — Q2 GAAP EPS $5.75 versus $1.82 expected, including a $53.4B gain, so "combined 150 billion of gains" across the two names from the Anthropic revaluation. On the operating side it is one of the three public cloud vendors where "value should continue to accrue to the infrastructure layer." And it is the reason July e-commerce looked terrible: "Amazon held its major Prime Day sales event in June rather than July, dragging down July e-commerce comparisons" — non-store retail −2.2% m/m, the single biggest driver of the retail-sales miss.
Full passage: premium transcript (PDF).
In short: Grouped with Meta and Alphabet as a real cash machine that could self-fund data centres from free cash flow until the scale broke it. "I'm not worried about Facebook and Alphabet or Amazon going bust… even if AI somehow goes to zero or nothing happens, I think it's manageable." The caveat he applies to all three: a chunk of reported net income is "other income" — mark-ups on stakes in Anthropic, OpenAI and SpaceX.
49:08It is quite extreme and it's a little railway-like. What do you make of CoreWeave's massive massive borrowings? — Yes, it's heavily indebted. There is in every cycle one or two or maybe a handful of outliers. Look, I'm not worried about Facebook and Alphabet or Amazon going bust. — Yeah.
In short: Named in the hyperscaler four: "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 moat is the barrier to entry itself — "the amount of money it takes to be a hyperscaler is insane and that expenditure itself is a moat."
Amazon is in the same defended category. AWS is one of the handful of places an AI company can physically run its models, and retail plus advertising provide revenue that has nothing to do with whether the AI boom continues.
Eisman's summary of the group is deliberately unromantic: "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." That is the whole claim — not that the returns on all this spending will be good, but that if the AI layer above them stumbles, these companies still have something left. The host's own reading of why Eisman prefers them is worth separating out as commentary rather than Eisman's words: moats, diversified revenue, and the balance-sheet strength to keep spending through an uncertain stretch.
9:42The amount of money it takes to be a hyperscaler is insane and that expenditure itself is a moat. 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. — So, Steve is much more inclined to look at the hyperscalers if you're going to play at all and it really comes down to three main reasons.
In short: The third named "hyperscaler winner." Also the historical template for the whole upstart segment: "Amazon conquered retail. Bezos started with a business model of disrupting Barnes & Noble… by mailing books from a garage. He funded growth with massive capex and never cared about earnings… And that risk paid off."
Amazon (through AWS) is the third named hyperscaler winner, on the same logic: a franchise business renting out computing capacity, protected by the enormous cost of building it, and exposed indirectly to the health of the AI labs.
Amazon also does double duty in this episode as the historical proof that upstarts can win outright. Bezos attacked Barnes & Noble by mailing books from a garage, spent heavily, ignored profits for years, and investors who backed that vision were rewarded — which, Eisman argues, is precisely why investors now reflexively pile into every new "disruptor" IPO, sometimes on the wrong 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: "I had Amazon before he bought Amazon" — disclosed in the overlap list as one of his own continuing positions Ackman later joined; no new argument made for it 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: Used twice as an illustration, not as a call. As a growth-to-value case: "When I started writing about stocks, in February 2016, Amazon ($AMZN) and Netflix ($NFLX) were still seen as high-growth stocks and way too expensive. Today, they are considered growth at a reasonable price." And as a founder case: "Amazon with Jeff Bezos." No position or view disclosed.
In short: Context only — listed with Apple, Microsoft and Netflix as the very-low-cost-basis megacaps whose holders won't sell: "you need to take your cost out and take some profit out."
34:12— Yeah. — I'm going to give you two instances where like if you look today at the big companies, Amazon, Apple, Microsoft, you look at the big names, Netflix, that a lot of people have owned for a long long time. And they have really really low cost bases. And their argument is I don't want to pay the tax, so I don't want to sell any of it, which we don't think is a smart move.
In short: The most recent call to have played out: highlighted at the start of the year as a top pick on the argument that "Amazon is being underestimated with AWS," backed by a dedicated "AWS is going to explode" video charting the reacceleration and the estimate revisions that would follow. "Amazon now looks like it owns a complete juggernaut" — +12% in the past month, +22% YTD, and on a ~$170 cost basis it has moved ~$100 above it: "a massive position, $200,000 with $77,000 in the green." Used as evidence for the pattern rather than re-argued.
Amazon is the most recent of his calls to pay off. At the start of the year he argued the market was underrating AWS, Amazon's business renting computing power to other companies, and made a dedicated video predicting the growth rate would accelerate and force analysts to raise their forecasts. That is what happened.
The stock is up 12% in the past month and 22% for the year; against a cost basis of about $170 it has gained roughly $100 a share, making it a $200,000 position with $77,000 of profit. He raises it here as evidence for the method rather than to re-argue it — the same shape as Microsoft, ASML and Google before it: a high-quality, fast-growing business that was being discounted for a reason he judged didn't hold up.
5:15And being able to correctly understand the bear case of the company and why it doesn't make sense. Now finally, the last one that I'll highlight is Amazon. This is one that I highlighted just at the beginning of this year as one of my top picks. I said that Amazon is being underestimated with AWS.
In short: The second half of the paper-gains problem — Amazon's Anthropic stake produced the same kind of non-operating unrealized gains as Alphabet's, so "the reported GAAP EPS for Alphabet and Amazon have the potential to decline in the first and second quarter of 2027." Operationally it was strong: AWS accelerated 28% → 37%, "a massive 900 basis point improvement, the highest growth rate since COVID," with margins +1%, so the stock rallied 15% despite guiding revenue and operating income below consensus. Jeff Bezos sold another $4.07B (15M shares) on Aug 3.
Amazon's cloud business had its best quarter since the pandemic, accelerating from 28% to 37% growth — a nine-percentage-point jump — with margins improving. That was enough for the stock to rally 15% even though the company guided both revenue and profit below expectations.
The catch is the same one as Alphabet's. Amazon holds a stake in Anthropic, whose valuation went from $183 billion to $965 billion during the year, and it books that increase as profit even though it never sold anything. For those gains to repeat, Anthropic must keep raising money at ever-higher valuations. If it doesn't, reported earnings fall in 2027 without anything changing in the actual business. Jeff Bezos also sold another $4 billion of stock in early August.
Full passage: premium transcript (PDF).
In short: Kevin Simpson's final trade: "Amazon — they're demonstrating how to make a return on their investments." Also named by Talkington (with Google) as confirming Microsoft's cloud-monetization read.
In short: Dean cites Amazon on both sides: as evidence frontier-model pricing had got too expensive ("companies like Uber and whatnot and Amazon pulling back on AI spending… people are blowing through their budgets in a span of a month"), and as one of the hyperscalers — AWS — that hosts the cheaper open-weight replacement and captures GPU rental economics instead.
4:32Essentially we believe this is bullish for enterprises because they can now spend a lot more with AI. You already heard talks about companies like Uber and whatnot and Amazon pulling back on AI spending because it was getting too pricey. People are blowing through their budgets in a span of a month and that's because they had to pay anthropic or chad significant amounts of capital to use their models.
In short: Named among the four hyperscalers driving the build-out, and in Brown's earnings-quality caveat: strip out Alphabet's and Amazon's investment gains (the write-ups of their stakes in Anthropic and other startups) and S&P earnings growth is still 28.8% next quarter. Also the alternative-chip counterexample in the neocloud argument ("Trainium"). No individual call.
In short: Her worked example of fake earnings quality: an initial $8B investment in Anthropic now marked north of $75B, and roughly $16B of a ~$30B reported quarter was that appreciation — "half of the purported growth" was a bookkeeping item with nothing to do with paying salaries, funding capex or servicing debt. On top of that it is issuing debt into widening hyperscaler spreads with free cash flow now negative.
Amazon owns a large stake in the private AI company Anthropic. It paid about $8 billion for it; that stake is now valued at more than $75 billion on paper. Accounting rules make Amazon run that increase in value through its income statement as "other income," so it shows up inside reported earnings — even though Amazon hasn't sold a single share and hasn't received a dollar of cash. Pomboy's number: of roughly $30 billion of reported quarterly earnings, about $16 billion was that mark-up. Half the apparent growth was bookkeeping.
Why it matters: those paper gains can't pay salaries, fund the data-center buildout, or cover interest. And the real cash picture has flipped — Amazon and its peers, once the cash cows of the S&P, are now free-cash-flow negative and borrowing heavily, at a moment when lenders are charging them more (the extra yield, or "spread," on their bonds is widening as investors worry about overinvestment). If AI valuations cool, the same accounting that inflated earnings on the way up produces large write-downs on the way down.
14:31view. Do you look at it similarly, or do you have other concerns that I didn't list there? — No, well, there is one other concern, which I'll get to in a moment, but the main point was, I think this valuation side that you mentioned at the top. Take Amazon for example. They I think their initial investment into Anthropic was 8 billion.
In short: One of the infrastructure players the value accrues to, and it passed both halves of his test in the June quarter: "AWS saw revenue growth accelerate 9% from the March quarter to the June quarter, to 37% growth, but more importantly to me is the operating margins also expanded by a percent."
AWS is Amazon's cloud-computing arm — it rents out servers, storage and, increasingly, AI capacity. It is one of the "infrastructure players" Niles says the money ends up with, on the logic that if AI models become interchangeable, the profit moves to whoever owns the computers the models run on.
The June quarter passed his two-part test. Growth accelerated by 9 percentage points versus the March quarter, to 37% — the pace picked up rather than just staying high. And, in his words the more important half, "the operating margins also expanded by a percent." Selling more while keeping a bigger slice of each dollar is the combination that says demand is outrunning supply.
The plain reading: AWS is being paid more for capacity it is also selling more of, which is exactly what you would expect if the binding constraint in AI has moved from clever models to raw compute.
3:22AND WHEN YOU LOOK AT THE RESULTS, AWS SAW REVENUE GROWTH ACCELERATE 9% FROM THE MARCH QUARTER TO THE JUNE QUARTER, THE 37% GROWTH, BUT MORE IMPORTANTLY TO ME IS THE OPERATING MARGINS ALSO EXPANDED BY A PERCENT. YOU LOOK AT AZURE, SIMILAR SITUATION WHERE THE GROWTH RATE FOR THAT ACCELERATED BY ABOUT 4% TO 43%, AND THE MARGINS EXPANDED BY A PERCENT.
In short: The centerpiece. AWS revenue "grew 36.7% year-over-year to 42.2 billion" — its fifth straight accelerating quarter and fastest in 18 — to a $169B run-rate, backlog +154% to $496B, operating income $16.6B at ~39% margins, with demand visibility "through at least 2028" and Jassy's first-ever trillion-dollar-revenue claim ("I believe Andy Jassy, he has the numbers to back this up"). Stock $285, +23% in five sessions, past a $3T cap; his Story Fund position is now $207k, +$82k, deliberately overweight alongside Google.
Amazon's cloud business, AWS, rents computing power and storage to other companies. For years the market worried that Amazon was pouring enormous sums into data centres with nothing to show for it. This quarter answered that: AWS revenue grew 36.7% from a year ago to $42.2 billion — the fifth quarter in a row where growth got faster, and the fastest in four and a half years. Annualised, that one division now runs at $169 billion a year: more than three times Netflix's entire revenue, about 70% bigger than all of Tesla, and roughly half of Berkshire Hathaway — which, at this pace, it passes around 2029.
Two details matter more than the headline. First, the backlog — money customers have already contractually committed to spend — grew 154% to $496 billion, faster than revenue itself, and Amazon says the capacity being built through 2028 is already spoken for. That is demand you can see two years out rather than forecast. Second, it is getting more profitable while growing this fast: $16.6 billion of operating profit at roughly 39% margins, helped by its own custom chips and networking gear. Growth, visibility and profitability all beat at once, which is why CEO Andy Jassy's claim that AWS becomes a trillion-dollar-revenue business gets taken seriously.
Carlson called this reacceleration in September 2025 and says he still lowballed it — "I underestimated it to a huge extent." Amazon is now $207,000 in his Story Fund, up $82,000, and deliberately one of his two overweight positions alongside Google.
5:56I underestimated it to a huge extent. The rate at which this behemoth is growing is actually incredible. If you take a minute to really look at the numbers here, this last quarter outlines a story that's so convincingly good that even the biggest bears are having trouble here. AWS's revenue grew 36.7% year-over-year to 42.2 billion.
In short: The market forgave the spend because the engine is compounding: AWS +37% to $42.4B, a fifth straight quarter of acceleration, generating a fifth of revenue and most of operating profit. 2026 capex raised $200B → $220B, sending trailing-12-month FCF to −$7.6B, "but because the operating profit growth is so strong, people ignored it." Shares +15% at the open, the biggest intraday gain since November 2022. Jassy: investment moderates over time — sites last 30 years, equipment 5-7.
Amazon raised its 2026 data-center budget from $200 billion to $220 billion, and its spare cash over the past year swung to negative $7.6 billion. Normally the market punishes that. It didn't — the stock rose 15%, its best day since November 2022.
The reason is AWS, its cloud division: revenue grew 37% and, crucially, has accelerated for five straight quarters. AWS is about a fifth of Amazon's sales but most of its profit, so as long as it keeps speeding up, investors treat the spending as growth rather than waste. CEO Andy Jassy's framing is that the spending is front-loaded because Amazon is building sites from scratch — the buildings last 30 years even though the servers inside are replaced every five to seven.
Full passage: premium transcript (PDF).
In short: The main story. Q2 FY26 revenue +20% Y/Y to $200.6B (a $4.2B beat), gross margin 52%, operating margin 14% (+2pp) — and AWS +37% to $42.2B, the fastest growth in 18 quarters, at a 39% operating margin (+6pp Y/Y; still +520 bps excluding a $600M energy-contract accounting gain). Backlog $496B; AI and custom-chip businesses each passed a $25B annual run rate growing triple digits (chips were above $20B only three months ago). "Accelerating growth and expanding margins at the same time is the quarter's defining result." Retail held up — North America +16%, International +15%, worldwide paid units +17%, NA margin flat as fuel and transportation offset fulfilment gains and a $600M tariff refund — and advertising +26% to $19.8B, near an $80B run rate. The bill: 2026 CapEx raised from $200B to $220B (mostly higher memory costs), trailing CapEx $169B (+64%), and TTM free cash flow −$7.6B against TTM operating cash flow of $161.4B (+33%) — capital spending now exceeds the cash the business generates. Net profit included a $53.4B non-operating gain, chiefly the Anthropic markup. Balance sheet: $123B cash and securities vs $129B long-term debt. Q3 guides to +9–12% revenue (Prime Day shifting into Q2 costs nearly four points, FX another 80 bps) with operating income ~$24.5B, +40% at the midpoint. Jassy's defence is the two-bucket return math (servers break even in <3 years against ≥5-year contracts; data centres monetize ~2 years out but run 30+ years across five or six server generations), most of 2027 capacity already reserved with commitments into 2028, and the claim that "the margins and returns in AI [are] tracking what we saw with Core at the same point of evolution, actually a little ahead." Moving up the stack: Bedrock customers spent more in Q2 than all previous quarters combined, AgentCore added payments/web search/deterministic controls, Amazon Quick runs autonomous workflows, Kiro tripled sequentially, Continuum remediates vulnerabilities. Watch items: Trainium as a possible merchant chip sold outside AWS, and grocery (same-day perishables in 2,300 US cities, monthly active perishables customers +50% YTD, perishable orders averaging 3x the units). The risk in one line: "If demand, pricing, or utilization disappoints, that operating leverage works in reverse." A disclosed author holding. (Analysis, not a stance call.)
Amazon will spend $220 billion this year building data centres and buying computers — $20 billion more than it said three months ago — and its CEO says it still won't be enough to serve everyone who wants capacity, this year or next. The reason he can say that with a straight face is AWS, the cloud-computing arm: it grew 37% last quarter, the fastest in four and a half years, while also lifting its profit margin to 39%. Growing faster and earning more at the same time, at that size, is rare, and it's the quarter's key fact.
The cost shows up in cash. Over the past year Amazon generated $161 billion of cash from operations but spent $169 billion on buildings and equipment, so free cash flow — the money left over — is negative $7.6 billion. Reported profit looks enormous, but $53.4 billion of it is a paper gain on Amazon's stake in Anthropic, which was revalued after Anthropic raised money in May at a $965 billion valuation (up from $380 billion in February). That is a mark-up, not cash coming in the door.
So why isn't this reckless? Jassy split the spending in two. Computers and networking gear are bought only a few months before they're switched on, once customers have already committed — and they pay for themselves in under three years while the contracts run at least five. Data centres are built about two years before they earn anything, but they last 30-plus years and will house five or six generations of computers. Most of AWS's 2027 capacity is already reserved, with bookings reaching into 2028. In other words: spending early against orders already in hand, not building on hope.
His bigger claim is about where AI demand goes next. Today it's "barbelled" — big AI labs and a few breakout apps at one end, small niche business uses at the other. The fat middle is all the ordinary corporate software that runs the world and barely uses AI yet. Jassy thinks that middle becomes the biggest slice of all, which would keep this buildout busy for years. Alongside it, Amazon is climbing up from renting raw computing into selling the AI tools themselves (an agent that writes code, one that fixes security holes, one that runs errands across your email and calendar), considering selling its own Trainium chips to companies with their own data centres, and pushing fresh groceries into 2,300 cities to make people shop more often.
The honest risk is stated at the end: Amazon is spending years ahead of demand for assets that could earn for decades — "if demand, pricing, or utilization disappoints, that operating leverage works in reverse." The author owns the stock; this is analysis, not a recommendation.
In short: A passing reference, used as the second exhibit for winner-take-most rather than as a call: "what about in e-commerce? That's just Amazon." Cited as the category whose winner-count is one.
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: Used admiringly as the real-world compounding "snowball" — 1,512 weekly closes since May 1997 trace almost the same power curve as bitcoin (at 50% of elapsed time the price was only 5% of today's) — and then explicitly qualified: AI data-center capital spending is rising dramatically, it "has just about transitioned from free-cash-flow positive to negative," and it has begun large-scale borrowing. "Will these changes conspire to degrade the rate of its actual future financial progress or its long-term valuation multiple (from the current-year P/E ratio of 36x)?" One of the AI data-center hyperscalers Fredrik Tjernstrom's May fundholder letter was "essentially a sell recommendation" on.
Amazon appears twice in this commentary, and the two appearances point in opposite directions.
First, as the proof of the compounding argument: 1,512 weekly closing prices since 1997 trace almost exactly the same curve as bitcoin and as a snowball rolling downhill. Halfway through its life as a public company, the stock was at 5% of today's price. Anyone who judged it by price action along the way would have sold repeatedly.
Then the caveat, which is the actual investment view. Amazon's spending on AI data centers is rising dramatically, it has just about crossed from generating free cash flow to consuming it, and it has started borrowing at large scale — all departures from the capital discipline that made it a compounder. The firm asks directly whether that degrades both its future growth and the 36x earnings multiple it currently carries. It sits inside the Magnificent Seven that this commentary values at 150 times run-rate free cash flow with $108 billion of annual stock compensation deliberately counted as the expense it is — and inside the group of AI data-center hyperscalers on which colleague Fredrik Tjernstrom's May letter was, in the firm's words, essentially a sell recommendation.
Full passage: premium transcript (PDF).
In short: A $170,000 position, +$16k on the open. AWS — "the thing they're spending all that capex money on" — grew 37% to $42 billion in a quarter, above expectations and at rising margins, with order fulfilment visibility showing demand increases as far out as 2028. "This is Amazon signaling to the market that they've struck gold." The stock ran +4% on the day then another 9–10% after hours.
Amazon's headline profit number is noisy because it includes gains and losses on stakes it owns in other companies, so Carlson ignores it and looks at the operating business. There, Amazon beat on both its own earnings and revenue — and AWS, the cloud division that all the AI spending is aimed at, grew 37% to $42 billion in a single quarter. Crucially it grew while getting more profitable, which is the opposite of what you'd expect from a business shovelling money into new capacity.
Amazon also indicated demand is booked out years ahead — order visibility showing increases as far out as 2028. That combination (fast growth, rising margins, multi-year demand) is what he means by "they've struck gold." The market agreed: up 4% on the day and another ~9–10% after hours. It's a $170,000 position for him.
7:33" And then Amazon reports their earnings. Now, we can't really look at the EPS number. A lot of that is equity stakes. So, that's something that we'll ignore for now. But Amazon organically beat on their earnings per share. More importantly, they beat on their revenue. And AWS, that whole cloud thing, the thing they're spending all that capex money on, it grew by 37%.
In short: "Amazon was strong… the market liked Amazon's results. EPS of 575 was up huge versus last year. And the most important data point for Amazon remains Amazon Web Services where revenue growth jumped to 37%. These results are kind of similar to what Microsoft reported" — stock +7% after hours. Also named as one of the four hyperscalers with "real businesses here."
Amazon's retail business is large and slow; the part that moves the stock is Amazon Web Services, its cloud computing arm. AWS revenue growth "jumped to 37%," which is the same signal Microsoft gave with Azure at 43%, and profit per share was "up huge." The stock rose more than 7% after hours.
In this episode's framework Amazon is a hyperscaler with a real moat — the cost of building at this scale is itself the barrier — and, crucially, one whose cloud growth is accelerating rather than merely absorbing capital. That is the distinction he draws between the companies whose AI spending is producing revenue and those, like Meta, where it is only producing costs.
19:43Missed expectations and China sales amounted to 18.8 billion. Also a disappointment. Apple shares fell around 4% after hours. By contrast, the market liked Amazon's results. APS of 575 was up huge versus last year. And the most important data point for Amazon remains Amazon Web Services where revenue growth jumped to 37%.
In short: Luria names it as one of "the three big hyperscalers" (with Microsoft and Google) that buy the chips and provide compute — a layer of the value chain that "adds just as much value, if not more, if the model is open source." Eisman's closing: the moat, if there is one, comes from Microsoft/Google/Meta/Amazon building out the data centers. Referenced as a layer, not rated.
12:01There's the companies that buy those chips to provide compute, primarily the three big hyperscalers, Microsoft, Amazon, and Google. And then there's the model companies. There's a lot of value being created throughout. And I'll point you to one important data point to show that which is the cumulative run rate of OpenAI and Anthropic right now is clearly above $75 billion in revenue.
In short: Q&A: "wait for sell-the-news first from Amazon, Microsoft this week, like Google" — capex will likely have gone up. Per Niles: AWS has a high bar after Google Cloud's +82%, higher oil pressures e-commerce logistics and consumer purchasing power, and moving the four-day Prime event to late June borrows Q3 demand into Q2. (Also the war's corporate casualty — the IRGC claims it destroyed Amazon's central data-center facility in Bahrain.)
Amazon reports this week and Singh's advice is to wait for the reaction, not front-run it: like Google, Amazon will probably announce it is spending more on AI data centres, and the market has just shown it punishes that. He expects a possible "sell the news" drop — buy after, not before.
The specific worries: AWS has a high bar to clear now that Google's cloud grew 82%; expensive oil raises delivery and logistics costs and squeezes shoppers; and moving the four-day Prime event from July into late June pulled sales forward, flattering last quarter at this quarter's expense. (Grimly, Amazon is also a casualty of the war — Iran's IRGC claims it destroyed Amazon's main data centre in Bahrain.)
Full passage: premium transcript (PDF).
In short: Two roles: the "Amazon effect" that made everyone give up on retail landlords ten years ago when it bought Whole Foods — "the pendulum has swung" back — and today an incremental demand source for logistics space ("Amazon is expanding"). Thesis input, not a stance on the stock.
37:33Everyone was like, "Oh my god, here comes big bad Amazon." We're not going to build new malls. So, the beautiful thing is, Canadian population has grown. US population has grown. They're pausing right now. But, if you're a Dollarama, if you're in a successful retailer and you're trying to find new space, well, okay, the Bay went bankrupt, Zeers went bankrupt, Sears is gone, right? So, there's no lowhanging fruit.
In short: Prime Video reached 4.2% of US TV time in April, helped by its new NBA package — the full-season-sports model Netflix explicitly declines to copy (Netflix buys tentpoles instead: MLB Opening Night, the Home Run Derby, the Field of Dreams game, five NFL games). Amazon is also named alongside ESPN as the destination Netflix is not trying to replicate. A disclosed author holding. (Referenced; not a stance call.)
In short: Raised CapEx "from 120 to like 200 billion"; part of the hyperscaler group whose incremental ROIC has fallen from ~40% to ~20% and is "moving toward 10%."
Amazon jacked up its planned AI spending from about $120B to roughly $200B. It's a profitable hyperscaler, so it has a cushion — but it's squarely in the group where each new dollar of spending is earning less and less.
Chanos's metric for the whole group: the incremental return has dropped from ~40% to ~20% and is heading toward 10%. At 10%, "you could earn that on treasuries" — which is when boards start questioning the spend.
32:14They're energy constrained. And this is going back I think it was like last October or November. All of a sudden though, are we starting to see the overbuild despite what we continue to hear about CapEx going higher, coming into the year Amazon raised their CapEx from 120 to like 200 billion or something like that.
In short: "The Amazons of the world, the Googles of the world" are "getting right into" nuclear because data centers need immense power that "doesn't exist right now" — "you can't just plug it into the grid. The lights would all dim."
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: Named in the mega-cap ramp (up ~3% with the hyperscalers) — one of the value opportunities capital is finding in the mega caps as it leaves memory/semis. No individual call this episode.
In short: Spending 100% of cash flows on data centers; got stuck with a $500M one-month token bill. Its recent $25B bond added 8bps to the 10-year and it now has to borrow in euros/yen/francs/pounds — it has saturated the US market, a sign of exhaustion.
Amazon is spending all of its cash flow on data centers and even got hit with a $500 million bill for one month of AI tokens. The tell he watches is in the bond market: Amazon's recent $25 billion bond was so large it nudged up the 10-year Treasury yield, and Amazon now has to borrow in euros, yen, Swiss francs and pounds because it has soaked up all the demand for its debt in the US. When a company has to scrape the bottom of every funding market, that's a sign of exhaustion near a top.
30:22— So, sorry to interrupt, but do you see this as late stage? Is this the people saying let's get out while the getting's still good? — Well, for sure in SK Hynix's case and certainly in SpaceX's case, the timing was good for them. They don't generate any money, any earnings.
In short: Link added to it — of the hyperscalers, "in a much better position of strength" across advertising, compute, retail and AWS, and the valuation "makes a lot of sense." The one she'd be buying more of (vs Meta, which she's not adding). More upside than Meta or the other Mag-7 spenders.
Among the big tech companies spending heavily on AI, Stephanie Link thinks Amazon is the best positioned — it has multiple strong engines (advertising, cloud computing/AWS, and retail) rather than one, and the stock's valuation "makes a lot of sense." She added to it and says it's the hyperscaler she'd keep buying, explicitly preferring it over Meta (which she's holding but not adding to). In short: more ways to win and a more reasonable price than its mega-cap peers.
In short: A commoditizer in the framework — "wants the AI unit cost to be extremely low… competitive and interchangeable," not focused on making its own flagship; monetizes AI through AWS infrastructure "as the hub themselves." One of the three (Meta, Google, Amazon) Carlson believes ultimately win the tug-of-war against the premium-model labs.
Amazon and Microsoft play the same role in Carlson's framework: they want AI models to be cheap, competitive and interchangeable, and they're not focused on building a flagship model of their own. That's because they make their money renting out computing power — Amazon through AWS, Microsoft through Azure — so they profit no matter whose model wins, and they benefit when models get cheaper and more of them run on their cloud.
Carlson groups them with Meta and Google as the "commoditizers" who, in his view, ultimately beat the premium-model companies (OpenAI, Anthropic) as AI models become nearly indistinguishable over time.
16:13They're mostly going to monetize through their massive infrastructure distribution and vertical integration. Amazon also wants the AI unit cost to be extremely low. They want the AI model to be competitive and interchangeable. They're not really focused on making their own. They just want everyone else's to be competitive and interchangeable.
In short: Named as a new Veolia collaboration partner — a recently announced deal to reduce water use at Amazon's data-center operations in Mississippi. A reference to the "rising" data-center water opportunity, not a stance on AMZN.
In short: Giroux: hyperscaler FCF winner 2030–32 — its low-cost ASICs (with Google's) are what Anthropic built the world's best LLM on; AWS Graviton named by Ellenbogen among the platforms LLMs shifted to. Black: 27x, reasonably priced among megacaps.
In short: AWS Bedrock likewise hosts Chinese models fully managed in the US cloud; GS also flags open-weight revenue-sharing on hyperscaler platforms (AWS Bedrock, Alibaba Bailian) as a potential high-margin monetization path for Chinese model companies.
Full passage: premium transcript (PDF).
In short: Guided 2026 AI capex to "200 plus billion" — one of the mega-caps whose ballooning capex budgets "started to scare people," even as they kept "racing to buy Nvidia chips."
Amazon guided 2026 AI capital spending to "$200 billion plus" — the largest of the mega-cap budgets, and one of the numbers that "started to scare people." It's another buyer in Eisman's frame: pouring enormous sums into the AI build-out (chips, data centers) with no guarantee those dollars earn an adequate return.
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: Cited (with Alphabet) as ~13.5% off its 52-wk high in the Mag-7 drawdown Wapner walked through — the debate being buy-the-dip vs a changing market makeup. No fresh single-name call.
In short: "An easy one" — doesn't face Meta's risks; already owns the logistics system, two-day delivery and Prime, and AWS "going forward will grow at an accelerated pace." "Amazon is spending more on capex than anyone… and all of it is going online and being immediately monetized." Another unique deal this month.
Amazon is Carlson's "easy" second July buy. Unlike Meta it isn't making a risky bet — it already owns the pieces that matter: a giant delivery network, two-day shipping, and Prime memberships. On top of that sits AWS, its cloud business, which he thinks is about to grow faster because demand for its services is so strong.
The reassuring part for him is that Amazon spends more on new capacity ("capex") than anyone, but "all of it is going online and being immediately monetized" — i.e. the moment it builds capacity, paying customers use it. So the spending translates straight into revenue rather than sitting idle, which is exactly what he wants to see.
15:37All of that equates to a very attractive company, an interesting one that again is selling at an 18 PE. Next up is an easy one. We also have Amazon. Amazon's a bit different than Meta because it doesn't face any of the same type of risks. Amazon has the benefit of already having their massive logistics system, already having the two-day delivery, having the Prime membership.
In short: AWS is named in Cerebras's risk disclosures as one of four significant customers — the key Western-enterprise name the bull case rests on — but AWS revenue does not begin until 2027. Also a disclosed App Economy Portfolio holding. Referenced/neutral.
In short: A logistics company mainly for itself — not yet expanding into FedEx/UPS's third-party business as feared; also cited as a place where "AI is clearly working" in the numbers.
29:52at these logistical companies FedEx UPS they throw Amazon in there but Amazon's really not it's a logistical company it's a logistics company but it's a logistics company for themselves right they're not expanding into FedEx's is business in a way that people would think yet so
In short: One of the hyperscalers behind the "lag-7": down on margin pressure (raw chip-input inflation into data centers), not fatally flawed — expectations were just too high. With Mag-7 margins at 25% unable to go much higher, a little downward pressure on margins/price. "This Mag-7 pullback is a setup for the next leg higher." (AWS grew 28% last quarter; ~24×, per Link.)
The "Mag 7" are the seven mega-cap tech names; this year they've lagged the market, so wags call them the "lag 7." Lebenthal explains why they're down and argues it's not alarming: they're spending heavily on AI chips for their data centers, and the price of those chips has jumped (input inflation), which squeezes their profit margins a bit. On top of that, these companies already run at very high (~25%) margins that simply can't climb much further — so a little margin give-back, plus expectations that had gotten too high, is enough to knock the share prices down.
His conclusion: Amazon and Alphabet aren't "fatally flawed" — the business engines (AWS grew 28%, Google Cloud 63%) are fine — they just got ahead of themselves. So he treats the pullback as healthy: "a setup for the next leg higher" rather than the start of trouble. (He flags one fair caveat raised on the desk — some of the group's reported earnings were flattered by gains on their private stakes in OpenAI and Anthropic — but says even stripping that out leaves ~25% real operating-earnings growth, which is strong.)
In short: Named alongside Meta as one of his "two favorite picks this year" ("I've highlighted Amazon and Meta as my two favorite picks this year"). Cited in passing here, not re-argued — the conviction carries over from his portfolio updates.
Amazon gets only a passing mention here, but an important one: Carlson names it, alongside Meta, as one of his "two favorite picks this year." He doesn't re-argue the case in this video — the detailed reasoning (AWS reaccelerating, advertising scaling, denser logistics, a strengthening moat) lives in his portfolio updates — but the conviction is clear and consistent.
28:05back here. Hopefully CNBC will do more of this. Now, moving on, we get to Meta with Mark Zuckerberg. Meta stock is currently down on the year. This is one that's one of my big buys for 2026. So, I've highlighted Amazon and Meta as my two favorite picks this year. And Meta makes sense for a couple different
In short: Amazon at 24x with AWS growing 28% last quarter — one of the reasonable Mag-7 multiples she likes against the cloud growth. Same FCF-burn / no-operating-leverage caveat applies across the cohort.
Same screen as Microsoft. Amazon trades around 24 times earnings while AWS, its cloud-computing business, grew 28% last quarter. Link sees that as a reasonable price for the growth you're buying. The same group-wide warning holds — heavy AI build-out spending means free cash flow can turn negative and margins won't expand much until the investment wave slows.
In short: "An eh" — middle of the road, a two out of three; lots of worse charts and better ones, no strong opinion. A name to play with options (a straddle) rather than take a directional view.
Amazon's chart is "an eh" — middle of the road, neither a strong buy nor a clear sell ("a two out of three"). Sohn has no strong opinion either way.
His practical suggestion is to express that uncertainty with options rather than buying or shorting the stock — for example a straddle (a bet that pays off if the stock makes a big move in either direction), or simply buying puts if you're skeptical and calls if you think it bounces back toward the middle of its range.
6:32AMZN (Amazon) — I don't think this chart looks so great. There are a lot worse charts and a lot better; it's pretty middle of the road, a two out of three. No strong opinion — an "eh." A similar one to play with options (a straddle; buy puts if skeptical, calls if you think it rebounds testing the middle of the range).
In short: "Still interesting long-term," add gradually on the Mag7 weakness: outspent everyone (~$44B capex in a quarter, ~$200B run-rate) and AWS margins slipped 140 bps to 37.7%, but AWS re-accelerated to +28% (~$37.6B) and the market gives no credit for the hidden ~$20B custom-silicon run-rate (Trainium/Graviton/Inferentia).
Full passage: premium transcript (PDF).
In short: Down 2.7% as the Apple-price-hike news hit the whole Mag-7 — "investors woke up genuinely concerned about this entire cycle." A holding caught in the broad AI-cost selloff (also notes "there's things Amazon does I don't like").
6:48We have Amazon reacting to this as well. It's down 2.7%. We have Microsoft, which is already a relatively cheap stock based on its growth and its quality, down 3.72%. We have Meta down 2.21%. It seems as though investors woke up genuinely concerned about this entire cycle of spending endless amounts of money, and now they're seeing it impact consumers directly.
In short: Organic revenue ~13%, EPS ~13–14%. Agrees with the "strengthened / high conviction" read despite competitors on every front (Walmart, streaming, Shopify, DoorDash/Uber last-mile) — AWS reclamation, advertising scale, denser logistics, Prime and AI infrastructure all reinforce the ecosystem.
Amazon faces competitors on every front — Walmart in groceries, streaming rivals against Prime Video, Shopify in retail, DoorDash/Uber in last-mile delivery — yet Carlson agrees its overall moat widened this year. The cloud business (AWS) is reaccelerating, advertising keeps scaling, its delivery network is getting denser (which is hard to copy), Prime membership keeps growing, and its AI infrastructure reinforces the whole ecosystem. Growing roughly 13–14%, it's a core holding he's comfortable owning.
In short: One of the hyperscalers "all down on Thursday" while the Micron-led rally stayed confined to chips and power — the capital-intensity/no-moats worry weighing on the group.
9:46However, and perhaps more importantly, the rally on Thursday was confined largely to chips and power related companies like GE Vernova. The hyperscalers, Amazon, Meta, Oracle, Google, and Microsoft were all down on Thursday. Apple also was down as it raised prices on its MacBook and iPad because of higher memory chip prices.
In short: Link: the only Mag-7 name she still owns — staying in the portfolio (probably smaller) while she trims Meta and avoids the other hyperscalers.
In short: Passing reference — scaling streaming advertising and a direct contender for the connected-TV living room Roku sits in. A disclosed author holding.
In short: Seth Klarman's biggest position — the value-investor counterpoint Amber presses him with; he concedes it's hard to argue there's no value when Klarman/Baupost owns it, though it's "more expensive than Meta at 17×."
10:46The counterparties to NVIDIA are a handful of companies that are dependent on that. So let's make your point. Seth Klarman owns, his biggest position is Amazon. Seth Klarman is one, we talk about him in our book, How to Listen When Markets Speak, value investor. My favorite value investor all time is David Einhorn, Charlie Munger. We sat down with Charlie Munger in the book, and Seth. And so, yeah, you're, that's why you're so good with the show and you create a great debate because how can I possibly make the case that
In short: Reportedly the "snitch" that flagged the Anthropic jailbreak method to the government — "maybe because they invest in OpenAI." Also one of the US top-5 in the Europe-vs-US comparison.
5:14The lesson here is that it's not a great business strategy to go to war with the United States government. Now, how did the US government magically become aware of a jailbreak methodology? Because Amazon told them. And why would Amazon snitch? Maybe because they invest in OpenAI.
In short: Grouped with Microsoft & Meta as a compounding machine whose "core economic engine" surpasses any passing cultural problem.
16:31Over the past 20 years, all these companies like Microsoft, Amazon, and Meta have gone through different periods of different cultures, different struggles, and different challenges. If you were to sell the stock every time it went into some challenging environment, you would sell it at precisely the lows every single time.
In short: Passing reference — SpaceX's post-IPO market cap (past $2.5T) was "nearly as much as Amazon." A disclosed author holding.
In short: In the hyperscaler debt/capex spree (raising ~$10B-equiv in a record C$ bond) headed past $1T next year — and the tokenomics tell: quietly removed its public token leaderboard as inference bills surprised higher.
Full passage: premium transcript (PDF).
In short: Incumbent Spotify out-competed; also invoked via Jeff Bezos — the "Bezos algorithm" (sacrifice short-term monetization for long-term dominance) and Bezos's word-of-mouth quote.
2:51If you don't believe that digital applications like Duolingo have a moat, then what is the moat of Spotify? And you have to answer the question as to how Spotify got over 60, 70% market share in the audio streaming industry despite being against larger incumbent players like Apple, like Amazon, like YouTube, and still dominated that industry despite having little less cash.
In short: In the hyperscaler-capex tally now headed past $1T by 2028 — the number whose downward revision (via slowing cloud-revenue growth) would unwind the whole trade.
43:46tech out of the S&P 500 right it's really barely moved on the year so you have a hyper concentration in basically one trade SpaceX is now basically an AI company right as well so I am a little bit concerned you know that with hyperscaler capex across Amazon, Google, Meta, Microsoft and Oracle now expected to be over a trillion by 2028 if that number were to come down or cloud revenue were to come down because the cloud backlog is one of the things that's driven this
In short: Same forced-selling group — one of the mega-cap index names passive funds have to trim to raise the cash the SpaceX listing pulls out of the market.
0:19You have ample opportunity to evaluate businesses. So, what one of the things we've been thinking a lot about is Newton's third law uh uh physics, which is for every action there's an equal and an opposite reaction. So, yes, SpaceX is going public, but where is the capital going to come from? And your point on the passive indexes, they're going to have to sell the largest constituents, which make up uh you know, are made up by names like Google and Meta and Microsoft and Amazon and Nvidia.
In short: 14% off highs, bottom of its 5-yr P/E range; "a less expensive SpaceX" — Leo/Starlink-like satellites, Zoox robotaxi, robotics, logistics, ads, Prime Video. Sees it above $300. Buy.
Amazon is the online-retail and cloud-computing giant. It's only 14% off its high and, on a price-to-earnings basis, sits at the bottom of its five-year range. Reported cash flow looks weak right now, but that's because it's spending heavily to build for the future.
His pitch is that Amazon is "a less expensive SpaceX" — a single company with a stack of moonshot businesses: Leo (its Starlink-style satellite internet), Zoox robotaxis, a giant robotics operation, logistics, retail, advertising and Prime Video. So many ways to win that he sees the stock above $300 and calls it a buy — and the reason he'd rather own it than chase the actual SpaceX IPO.
12:17It is only 14% off of its highs. It's trading in the middle of its 52- week range. Amazon also on a historical valuation looks very cheap based off of its PE ratio. It's at the very bottom range over the past 5 years. The cash flows are non-existent. Amazon's going into the red based on its free cash flow.
In short: Pitched AWS in 2013 ("won the war before it started," a 7-year lead, "got it for free") — the canonical mega-S-curve, ecosystem + scale moat.
Amazon is his textbook "mega-S-curve" win. Back in 2013 he pitched it for AWS — its cloud-computing arm — when that business was a hidden line item the market ignored; he said "the bulls have no idea what they're sitting on," that Amazon had "won the war before it even started" with a seven-year head start, and that you were effectively getting AWS "for free" at the price.
The lasting moat is the combination of being first, becoming a platform others build on, and reaching such enormous scale (10× rivals) that no one could afford the R&D to catch up.
34:43and cost to acquire versus lifetime. It's the whole business model. And so almost all the companies I mentioned have Apple, they have all of these rolled into one. Um, so we can sometimes we can notice these things before the rest of the world. And one of our high points was we pitched Amazon for AWS at 2013 at the Robin Hood investors conference and we said the bulls have no idea what they're sitting on.
In short: The clearest squeeze — ~$200B of 2026 capex (+67% Y/Y) with free cash flow collapsing to ~$1B (from ~$38B in 2024) even as operating CF hit ~$149B and AWS reaccelerated to ~28%. "Growth and free cash flow pulling in opposite directions."
Amazon is the online retail giant plus AWS, its rented-computing arm that powers much of the internet. AWS just sped back up to about 28% growth — its fastest in almost four years — so demand is strong.
The catch is the bill: Amazon plans to spend roughly $200 billion this year building AI capacity, up about 67% from last year. The business still generates plenty of cash (about $149 billion from operations), but after paying for all that construction, the leftover cash ("free cash flow" — what's left after the company funds both day-to-day operations and big physical investments) cratered to around $1 billion, down from about $38 billion in 2024. In the article's words, growth and free cash flow are "pulling in opposite directions" — Amazon is essentially plowing nearly everything it earns straight back into the AI buildout.
In short: The super-app comparison/competitor — Uber's Expedia tie-up is framed as the first move toward something that "competes structurally with Amazon Prime."
In short: Held in his Story Fund and the centerpiece of his rebuttal: AWS sold S3 storage — a commodity — at 30% operating margins by wrapping it in availability, security, compliance and tooling. Proof a commodity input can earn far-above-commodity pricing.
Amazon is the online-retail and cloud-computing giant, held in his Story Fund and the centrepiece of his argument that AI can still be very profitable. His proof: Amazon Web Services (AWS) sells plain online storage — about as basic a product as exists — yet grew it past $60 billion a year while keeping roughly 30 cents of profit on every dollar of sales, far above what a basic product should earn.
How? It didn't just sell raw storage; it wrapped it in security, reliability, global access, compliance and thousands of tools that make it easy and safe to use. The plain product is cheap and copyable, but the whole package around it isn't — and that package is what lets Amazon charge premium prices. He argues AI will work the same way: the model itself may be a commodity, but the company that packages and delivers it best will still make great money.
11:35Because we can look at other examples of how this worked with other companies. For example, if we look at Amazon, one of the biggest ways that Amazon makes money is through Amazon Web Services. Amazon Web Services existed long before artificial intelligence, and it was an incredibly good business before AI.
In short: The biggest spender in the arms race — $220B of capex this year (vs Google's $180B, Meta's $135B); cited as evidence the table stakes "have exploded" and capex keeps being revised up, not down.
49:55spending. I have to spend because if I don't, they may win and I be I may be out of business. The meta, by the way, the meta problem — is that — Google is spending 180 billion this year. And actually, I'm very proud of myself that I actually know these numbers now. And Amazon's spending 220. — The table stakes in this business have exploded.
In short: Exhibit of uneconomical AI spend — reportedly burned ~$500M on AI tokens in a single month, "realized oops," and shut it off.
Amazon shows up as proof that AI is expensive, not cheap. Companies pay for AI by the "token" (roughly, per chunk of text the model processes) — the more you use, the bigger the bill. Dowd says Amazon ran up about $500 million in a single month, "realized oops," and shut much of it off.
His point isn't about Amazon's stock; it's that even the giants are discovering AI doesn't pay for itself yet — a crack in the boom.
6:46using AI than perhaps outsourcing — yeah that's that the AI spend is so an so uneconomical at the moment that companies are are quickly uh shutting off their uh their token use. One company and I believe it was Amazon spent 500 million in one month and they realized oops we made a mistake and they're shutting that off.
In short: One of the five hyperscalers (with MSFT/GOOGL/META/ORCL) whose combined CapEx fell vs Q4 — his key signal that AI-CapEx growth is decelerating.
Amazon is cited only as one of the five cloud giants whose combined AI spending dropped quarter-over-quarter. Woo isn't judging the stock — he uses Amazon's spending as part of the evidence that the AI investment boom is starting to decelerate.
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: Named with Google as a data-center copper-demand driver — the "if they build it" scenario behind a physically impossible copper supply challenge.
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: Referenced via AWS — Snowflake's 5-year $6B commitment uses AWS Graviton CPUs and custom AI accelerators to lower compute cost and support Snowflake's 75% product gross margin. Also a disclosed App Economy Portfolio holding.
In short: The Starlink competitor — Amazon Leo (Kuiper) begins US service mid-2026 with 300+ satellites and Amazon's balance sheet behind it, a key pricing-pressure risk. Also App Economy's "Musk premium" foil: Amazon Leo "doesn't move AMZN." A disclosed author holding.
In short: A top phase-3 "buyer" and one of "my two top picks this year." A $188k Story Fund position; monetizes AI via AWS, Trainium, retail automation, ads, logistics. "In a great position."
Amazon is one of Carlson's two favorite picks of the year (the other is Meta). In his framework it's a "phase 3" buyer — a hyperscaler that spends heavily on AI now but will monetize it for 10–20 years across AWS cloud, its Trainium chips, retail automation, advertising, logistics and shopping assistants. It's a $188,000 position in his Story Fund.
The key idea: today Amazon's spending depresses its reported profits, which makes the scarcity-phase sellers look better by comparison. But it "owns the customer," so when AI compute gets cheap and plentiful, Amazon turns all that spending into durable, long-lived profit. "I think Amazon is in a great position."
14:11So, you're looking at a long lifetime of monetization. Amazon is doing this through AWS, Trainium, retail automation, ads, logistics, AI services, shopping assistance, and so on and so forth. There's so many ways that Amazon's monetizing AI. With Meta, we have ad efficiency, recommendations, AI systems, business messaging, content generation, wearables.
In short: Named (via AWS) as one of the hyperscale customers driving NVIDIA's new Hyperscale segment (+115% Y/Y to $37.9B). Also one of the author's disclosed App Economy Portfolio holdings.
In short: Trainium is "by far" the most underestimated custom silicon — aggressive design choices, and Amazon's Trainium runs one of only two functioning switched scale-up networks in the world (the other being Nvidia's).
Beyond its retail and cloud businesses, Amazon designs its own AI chip, "Trainium," as an alternative to buying Nvidia. Baker thinks Trainium is "by far" the most underestimated of the custom AI chips because Amazon made aggressive, ambitious design choices (where Google played it safe with its competing TPU).
The clincher: running the newest AI models (which split work across many specialist sub-models, called "mixture-of-experts") requires that special chip-networking he describes — and only two such working systems exist worldwide, Nvidia's and Amazon's Trainium. That scarcity is why he rates Amazon's silicon so highly.
16:51He's still going to be the dominant part of the market, but competition is coming. As you think about Trainium, TPUs, MTIA, which of these is most underestimated by the market? Where does the consensus get it wrong? Trainium by far. Trainium is going to be to 2026 especially in the second half of this year when Trainium 3 really ramps as TPUs were to 2025.
In short: The other Other Income offender: reported ~$16bn of Other Income in 1Q. A large Anthropic backer whose $100bn compute commitment feeds the same loop (invest → mark up the stake → book the gain as earnings while booking the cloud revenue). Evidence in the circularity critique, not a stance.
Amazon runs the same play as Google: it invested in Anthropic, gets to mark that stake up as the valuation climbs, and booked roughly $16 billion of "Other Income" from these kinds of gains in a single quarter — while Anthropic separately commits to spend $100 billion on Amazon's cloud. Investment goes in, the mark goes up, the gain shows up as profit, and the customer's spending shows up as revenue. Paulo's point is that when you strip out these self-referential paper gains, the underlying cash generation is far weaker than the headline earnings suggest.
In short: Historical analogy, not a current call — his first stock at Fidelity, bought in '98 when there were "only 100 million internet users and only 2 million e-commerce users… it doesn't even need to grow for this stock to be a buy"; the template for buying an L-curve early.
6:22— So how do you think about investing in an L-curve?
6:48And I said, "It doesn't even need to grow for this stock to be a buy." This one's moving faster. And the revenue growth that we're seeing at Anthropic — going from 100 million to a billion to 9 billion, and then it's already at 45 billion. It's going to be maybe 100 billion.
In short: Nice beat — EPS 2.78 vs 1.63 est, revenue 181.5B vs 177.3B est. AWS reaccelerated to +28% (vs 24% last qtr, 26% est); CEO cited >$225B in revenue commitments to train AWS's AI chip. Stock flipped from down after hours to +1% Thursday.
Amazon's quarter was a clean beat — earnings of $2.78 versus $1.63 expected — but the part that mattered was AWS, its cloud unit, reaccelerating to 28% growth (up from 24% the prior quarter). Management also said it now has over $225B in revenue commitments tied to its own AI training chips, a sign customers are betting on Amazon's AI hardware. That news flipped the stock from down to up on the day.
23:11Most importantly, Amazon Web Services Cloud unit revenue rose 28% versus last year. That is better than last quarter's 24% and better than the 26% estimate. On the call, the CEO said that the company now has over 225 billion in revenue commitments for training AWS's AI chip. And that news really excited investors and the stock moved from being down after hours to up 1% on Thursday.
In short: Held by two of the listed vehicles — Scottish Mortgage 4.0% and Markel 3.7%. Structural reference only; no stance.
In short: Appears three times without a view: #2 on both Top Buys tables, one of the eight quality-fund adds, and the comparison used for MercadoLibre's moat. No thesis or valuation offered.
In short: Hyperscaler contagion evidence: on Friday's euphoria, "hyperscaler bonds barely rallied (AMZN sold off along with the stock)" after being "annihilated on the capex guide" — yet retail piled straight back in (Vanda). Software weakness boomeranging back onto the Mag7.
In short: Bill — Mag-7 / concentration reference. Named in the FANG → Magnificent-Seven group whose top-10 dominance "is not going to get rearranged for the benefit of these investors." The over-owned cohort due to turn over.
21:16What I like to say is show me anywhere in the Old or New Testament of the Bible that God said everybody's supposed to make a lot of money on their stocks without having to even apply any serious thought to it. And the answer is it it it's not in there. It's not going to happen. uh all you know is something really bad is going to happen uh to stock market prices of the overowned most popular things and and that is coming that is coming.
In short: Valuation/playbook benchmark — Coupang is its Korean analog (minus the AWS kicker); the mid-2000s Amazon capex-inflection (reported margins lagging economic reality until scale was undeniable) is the template. AMZN trades ~3× sales vs CPNG's ~1×. Not a call on AMZN.
In short: Illustration only — no view expressed. Used twice as a stand-in for "a company that just reported": the incumbents' manual data factory when "the Amazon 10-K comes in", and the latency claim — "when Amazon does report their numbers at the close today… you're going to get it at 4:02 p.m., not 4:02 PM in three days from now."
30:23And so, for example, in that case, what takes them two days, we do in two minutes, so that when Amazon does report their numbers at the close today — they're not actually reporting today, but hypothetically — you're going to get it at, you know, 4:02 p.m., not 4:02 PM in three days from now, right? And so that's a key differentiation in terms of the actual infrastructure of this data that is, yes, in the public domain, but how you actually bake the cookie matters in terms of how it tastes.
In short: Named among the last-mile logistics platforms (with DoorDash, Uber, JD Logistics) "now orchestrating fulfillment in real-time," and as the AWS analogy for how monetization expands with usage as movement becomes programmable. A sector-illustration reference, not a call on AMZN.
In short: Cited as another survivor born in the 90s that "didn't build the infrastructure" — an asset-light winner, like Google. Same caveat applies now that the hyperscalers are the ones doing the capex.
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: Reference: AMZN priced a $13bn bond offering that morning — one more slug of "heavy" hyperscaler paper, illustrating the supply pressure on the AI-capex credit complex the note is about.
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.