CNBC Halftime Report — Debating the Road to S&P 8,000

The "contras" and the earnings-bubble rebuttal (1:24)

• Brown keeps a roster of commentators who are reliably wrong — "they have podcasts, so it's super convenient" — and uses them as a contrary indicator. His observation: the people who in 2024 complained the bull market was hyper-concentrated and driven by multiple expansion now complain about an earnings bubble, "which is hope. You can't win with these people… if that's the problem, that there's too much earnings, I'll take it."
• The data he answers with: Q2 growth around 50%, or 31% stripping out Alphabet's and Amazon's investment gains (they wrote up stakes in Anthropic and other startups, "not obviously repeatable"); on July 1 the estimate was 24.4%, so estimates are rising. For next quarter it's 28.8% ex those gains.
• The revision statistic is the centerpiece: "in the last 80 quarters — that's 20 years' worth of data — the average change in estimates for the first month of the quarter has been −1.9%. Right now it's +0.1%." Analysts normally lower the hurdle mid-quarter; instead "they have to. They're listening to management."
• And the quality check: "all 11 sectors are reporting year-over-year sales growth… we're talking about sales growth of plus 14% in the quarter. You can't fake revenue. You could fake earnings, but that's not what's happening." Tom Lee's version: organic EPS growth of 21% in Q2, 2027 estimates still rising, "we're dip buyers, we see the S&P reaching 7,900 to 8,000 this month."

Ethridge's timing objection — the pre-funded CapEx pull-forward (4:36)

• He doesn't disagree with the premise: as long as the four hyperscalers "throwing trillions of dollars at the problem trying to create as much compute as possible" keep reaffirming and raising their spending plans — which they all just did — "that is reason to feel bullish. It's obviously trickle-down economics… 2nd and 3rd order effects" (Caterpillar's best year ever because of Microsoft, Amazon, Google and Meta).
• The caveat is calendar-shaped: "I think that probably is through the end of this year… A lot of companies pre-funded a lot of their purchases for next year this year, because they'd rather pay today's elevated price versus next year's elevated price — the known known versus the unknown unknown." So the question is how much of the third-order spending (Caterpillar, Corning) "is going to slow down because the companies accelerated this year and they won't do it again next year."
• Brown's counter to the "it's all one theme" framing: "Is Allstate selling GPUs? Is MetLife selling GPUs?" — insurers are companies "using AI, turning it on their businesses, weaponizing it against the problems that they once had," so operating margins improve without any AI revenue at all. Ethridge concedes the theme is broad but insists the market cares about one theme trickling through everything it touches.

Baruch's three pillars — Warsh, the Strait, and the compute constraint (7:42)

• Pillar 1 — Warsh is misunderstood. Written to clients two weeks ago, ahead of the Fed meeting: expectations for hawkishness "were really offsides," and even now the market prices better than a 50% chance of a September hike. Wapner pushes back with the FT reporting Warsh is prepared to hike in September if inflation stays hot; Baruch's answer is that inflation "is not hot yet" (hotter than target, yes) and that the hawkish anecdotes "are going to become incrementally walked back… a tailwind into the midterms." He cites Powell's "oil price shocks are not to be controlled by monetary policy" precedent as a template for looking through the current oil rise.
• Pillar 2 — the Strait. "Trump does not want voters at the ballot box in the midterms with the war taking place and oil above $90," so "they have to engineer some sort of passage through the Strait of Hormuz, which is starting to show — the market is going to find that favorable. There's no better way to goose the economy."
• Pillar 3 — the market is compute-constrained. "Alphabet even said on their earnings call that they're going to purchase more compute while they build out their compute," and Meta is "leaning into being maybe a 2027 story, but they're going to sell compute." Two weeks ago with the SMH 25% off its highs, these were the three reasons to be bullish; the pillars haven't changed and "they are maybe the catalyst that gets 8,000."
• On the 30-year at 5.21 he's a skeptic of the breakout: as a trader he'd consider buying way-out-of-the-money 30-year puts as a tail hedge, but when respected bond people "get super excited about these rates and start calling for a 7% ten-year — the harder part of their career was in the middle of rates above 7 to 8% — usually that becomes the top in yields."

Brown: the labor market and the consumer aren't cracking (11:35)

• "Let's not gloss over why yields are where they are… the bigger picture is that the labor market is actually showing signs of re-acceleration. Despite all the talk about AI disrupting jobs growth or leading to mass layoffs — none of that is happening, none of it's in the data." He points to incomes in the bottom decile rising faster than the general population: "these are the things that will help people contend with temporarily higher energy prices." (Wapner's counter: ISM services employment came in at 47.4, a contraction.)
• On the consumer, he prefers management testimony to prints: Moynihan says "there's absolutely nothing going on in terms of delinquencies"; Jeremy Barnum at JPMorgan gets asked every quarter for the moment the consumer cracks and every quarter says it didn't; and Uber's Dara told Andrew Sorkin that riders aren't trading down from Uber Black and aren't reducing tips, across roughly a billion transactions a quarter. "We could sit and read non-farm payroll reports if we want, or we could listen to the people."

"A healthy bull market takes out its own trash" (14:43)

• Wapner: the market has looked past a 30-year at 5.21 and a more hawkish Fed (three dissents) "because of the earnings story. If you didn't have the durability of the earnings growth, the market would look far different." There were also deleveraging moments — the momentum trade unravelling in a blink, a hedge fund blowing up — "but the reason they weren't more upsetting is because everybody's falling back on what is a healthy market."
• Brown's formulation: "a healthy bull market takes out its own trash. We saw a momentary speculative fever take hold, predominantly focused on memory chip makers and a lot of it having taken place in Korea — and people paid for it. Those 2× leveraged single-stock ETFs got wiped out. People in certain options trades got wiped out. We had one notable hedge fund [with], I think, the biggest trading loss at a hedge fund ever in dollar terms. But the entire system didn't buckle."
• The detail that made it worse for the fund: it "was not only long memory, it was short software, and those shorts went against [them]" — IGV was 37% off its highs in April and has come nearly all the way back. Baruch's corroboration: healthcare is up 20% off the low over two months — "as momentum was getting hit, you're seeing the money go elsewhere."

Baruch buys Marvell after a 37% July (16:47)

• He waited for confirmation rather than catching the knife: "Monday was a great day for the market and then we didn't want to really buy the name just yet — obviously South Korea, let's see how that night goes and see if there's follow-through." Marvell was then up 14% on Tuesday and "we still bought it. We bought half as much as we were hoping for, which is a 1% add to our portfolios."
• The thesis: "they're emerging as the connectivity backbone with AI custom silicon, optical networking and AI memory solutions. CXL — Compute Express Link — allows hyperscalers to expand and pool memory outside of processors and works as a complement with high-bandwidth memory. I think they're going to be a huge leader as this AI build-out continues."

NVIDIA below the market multiple, and the circularity charge (17:43)

• Up 12% in five sessions, reporting August 26 as the last big-cap tech to print. Ethridge: he expected the rotation back "into the one that really should matter, because NVIDIA is trading now below the rest of the S&P 500 on a forward multiple — absolutely absurd when you consider it's the largest company in the world by market cap" and the company at the center of everything, with "like 90% of that [AI spend] flowing through NVIDIA." He's unsurprised it trades up into the print; "the surprise will be if it continues to go up right after."
• The bear case is circular financing — the Cisco-1999 analogy where the vendor funds its own customers. Brown's consistency test: "if you question that, you have to question the entire thing. Do we still feel really great about SK Hynix and Samsung, or do we not? If you believe in one, you have to believe in the other… it's like pulling the wrong Jenga piece out."
• And the practical framing: ten neoclouds are building data centers without their own cash flow (unlike Apple, Alphabet and Amazon), so they borrow — from Blackstone and the like. "If you're Jensen, do you want NVIDIA chips in there, or Trainium chips? They sort of have to go where the market's going." Plus NVIDIA is "the last of the mega-cap techs that's still massively free-cash-flow positive" at $48.5B, "so they still have the ability to lend out dollars in the form of structured loans."

SpaceX's lock-up opens — three positions, three answers (20:44)

• Morgan Brennan's numbers: available shares up more than 140% today to over 1.5 billion (from 629 million at the IPO), billions more unlocking into 2027 and 50% of shares in free float by January (Bernstein). Offsetting the supply: some indexes changed methodology before the IPO to allow faster inclusion, so "a bigger float will mean a bigger weighting, which in turn could mean more forced buying" — the Nasdaq-100's float adjustment comes in early September. Long-time holders say they'll hold and Ark has been buying the decline. And the day's news: the Terrafab Texas AI-semiconductor facility with Tesla, $16.8B for the initial phase, with combined SpaceX/Tesla chip demand "in excess of 1 terawatt of compute, significantly larger than the current global supply" — one reason CapEx is ramping.
• Baruch owns it: bought over the past week, "one of the things I just want to put away and forget about, and hope we see this thing find its story over the coming six months." Starlink is what everyone leans on and data centers in space "sounds like a terrific story, but the compute side is really where I think this is" — the Cursor acquisition and partnership, where they train their Colossus models.
• Brown says too hard: "I don't understand what the opportunity is even here after the share price has gotten cut in half." They raised $86B in the IPO — "the largest IPO raise anybody's ever seen on God's green earth" — and "have already spent like a fifth of it in a quarter, so they're going to need to raise capital again." On the float: "if you didn't like it with only 5% of the shares available to trade, you're going to love it at 12%." His structural read: "the IPO was engineered to perfection — the immediate inclusion into the indexes, the 5% float, the insiders being locked up and not allowed to even think about selling. This is the first click anybody's going to get."
• Ethridge wants a price: he was the pre-IPO skeptic ("I'm the rational person who would come on the desk and put cold water in people's face… I don't know that I want to buy a $2 trillion market cap on 5 billion in revenue and wait for that 5 billion to get to 100 billion and still have it overvalued"). Now: "I'd love to buy it under 100. I think I'm going to get a crack at it under 100 — depending on why it's under 100." Baruch: "115 was a level that I liked on the way down… I would add a little more below 100." Brown's needle: if you'd buy it at 115 and at 109, "we're totally insensitive to price, totally insensitive to valuation, and this is about the emotional need to belong to the story."

Ethridge sells Spotify — the AI-slop cost line (28:02)

• Down 18% ytd and 8.5% in a week after reporting higher costs tied to new AI-powered app features and guiding subscriber growth slightly light. "This was a much growthier company back in 2023 when I first bought it," and the reason he held was pricing power — repeated price increases without denting paid subscriber growth.
• What broke the thesis is the expense line, on two fronts: marketing spend to acquire the next customer ("kind of a concern"), and — the novel one — "operating expenses going up because AI slop on the platform has gotten so out of control. They talk about how much time and how many tools they've had to create in order to manage that AI slop getting onto the platform. And I think it only gets worse from here" as AI tooling proliferates. "As the OpEx goes up, that tends to be one big problem they have a hard time getting away from."
• Relative valuation closes it: 34× forward "when you could own something like a Netflix that's trading below the market at this point, and they kind of operate in the same space." Brown on the stock's history: "this was 800, then down to 400 — it's unbelievable, the violence with which people change their minds, positive and negative"; he bought "in the 140s because of the freak-out over the Joe Rogan experience… a bet that people aren't going to cut the cord just because they don't like one creator."

Joby as the spec — buying revenue to fund the CapEx (30:48)

• Up 6.5% after beating, reiterated neutral at Canaccord. Ethridge: "it is my only eVTOL play. I was in Archer as well — I got out of Archer, consolidated into Joby. I think this will be the first one to truly commercialize eVTOLs and they are way ahead of the field."
• The distinguishing asset is ordinary revenue: last year's acquisition of Blade — the helicopter shuttles from Manhattan to the airports and the Hamptons — "brought in $120 million-plus annual revenue, and revenue becomes really important when you're funding massive CapEx to launch this fleet of next-gen." Toyota adds a JV and an equity stake; he expects the air-taxi service flying in Dubai and Abu Dhabi by year end.
• The honest label: "I don't think this is the type of company you should invest in if you need earnings and dividends, because you ain't going to see either one of those for a very long time. But as a spec, this is one of my favorite plays."

Uber: an incredible report nobody believes (33:18)

• Wedbush to $91 after earnings. Brown: "I thought it was an incredible earnings report, and I said the day before the report don't worry about it, because nobody cares. The stock is down 6 out of the last 7 quarters after reporting. Doesn't matter if they beat, if they miss, if they guide up, if they guide down."
• The single unpriced variable: "nobody believes that the strategy Dara is pursuing on AVs is going to be competitive with Waymo and Cybercab. I do, but I am obviously in the minority, and that's why the stock trades at 15 times forward earnings. If anybody believed they'd be able to put together this coalition of third-party AV companies that would flood the platform with available rides and win, the stock would be $120 a share."
• What actually re-rates it, and it isn't a quarter: launching with NVIDIA; another ten OEMs rolling AV-ready cars off assembly lines; and financing — "some big private equity, private credit players that recognize, hey, this is a business, let's buy a fleet of AVs, we'll put them on the Uber app. The average AV on the app right now is 15 or 20 rides a day. That's a business. We take out the most expensive part, the human driver who has a take rate of 20–30%. When those things start to become real and not just things that are talked about in the conference call, this stock could be re-rated overnight. I just can't tell you when."

Best stocks: Block flagged for exit, Visa spotlighted (37:04)

• Block — talked about at 76, ran to 86, now pulling back after "a pretty good earnings quarter" with awkward expenses: "this is the company that reduced headcount by like 40%, and for some reason their expenses are rising. So this is what I mean by nobody really trusts Jack Dorsey for long." The rule: 81 was the breakout, it's retesting, "I would be very careful remaining long this trade… a close to end the week below 80, I would be just taking this one off the table."
• Visa — "one of the greatest growth stocks of all time, up 3,250% since its IPO. That is a 33× return… basically one of the best businesses in the world. The consumer trusts them, the merchant trusts them, and they are finding new ways to win regardless of what happens with the economy." Revenue +11%, earnings +11%; a golden cross on the chart; the May 2025 high at 375 unvisited until now — "when it gets through, there really are no natural sellers here. The path to 400 is clear and wide open," levered to travel, the global economy and employment.
• Ethridge (owner) supplies the durability argument: "why do we always talk about Apple's massive installed base but we never talk about the largest processor in the world?" — ~300 billion transactions a year, "literally in everybody's pocket the same way the iPhone is," not easily disruptible by AI, and third in blockchain patents behind Bank of America and Mastercard. "They'll continue to be sitting right between every single swipe transaction that happens forever… one that you continue to own forever in your portfolio rather than trading in and out of it."

XLE's two-sided options tape, and Berkshire into Saturday (41:20)

• Renick on XLE, with energy vying with tech for best sector ytd: more puts than calls by volume (11,000 vs under 8,000), "but we're also seeing a lot of put selling as well, meaning traders don't think the group will fall far." Two large bullish prints — ~$175k on a deep in-the-money 35-strike stock-replacement call expiring tomorrow, and $150k of 60-strike September calls (a 6% rally bet) — against heavy 55-strike September put volume needing a 5% drop.
• Baruch's energy positions: Exxon, Kinder Morgan, Cheniere. "Structurally, crude oil will have a higher floor, but we really like the pipelines, the midstream… they're going to continue to get paid on moving [it] into power. It's just under-invested, and the demand in the United States is going to be needed for power [and] AI."
• Berkshire reports Saturday. Brown: the stock has worked on the portfolio (Apple, Coca-Cola, American Express) and the insurance leg (Travelers, Chubb, Allstate, MetLife — "all of those charts look great"), plus "the railroad's on fire" and "one of the largest collections of utility assets in America. It's almost like everywhere you look, Berkshire is making money… you don't have to invest directly in AI to have benefited from AI." Baruch, who bought more right after the Alphabet-stake news broke: "the insurance business has been a huge compounder… we've been leaning into it on weakness. I like this from a defensive compounder standpoint."
• Final trades — Baruch: Amgen ("six growth drivers blowing it out of the water… you probably buy it a little bit lower"). Brown: Toast ("at a year-to-date high and I really like how it's setting up"). Ethridge's pick is garbled in the transcript.