Speaker · Updated
Brad Gerstner: latest market views
Altimeter Capital
Summary
Brad Gerstner is the founder and CEO of Altimeter Capital, a technology-focused hedge fund and venture capital firm known for investments in companies like Snowflake and Uber.
Chorus has 31 views from Brad Gerstner across 5 episodes on 4 shows since October 15, 2026. Their latest calls: bullish on Anthropic, NVIDIA, OpenAI and Space Exploration Technologies.
Where they stand
Their latest bullish or bearish call on each company.
What they talk about most
The latest views
Newest first. Every quote is checked against the episode's transcript.
AnthropicBullish
Anthropic IPO may price at the low end, but demand remains strong and Altimeter will be a major buyer.
Gerstner, an early Anthropic investor through Altimeter, says the market turmoil will make overall sentiment less euphoric and push the IPO toward the lower end of its range. He still sees a serious level of demand for the offering, and his enthusiasm for the company has not changed. He says Altimeter hopes to be one of the largest buyers.
“So it will probably... You know, this pushes it a little bit more to the lower end of the range wherever it was going to price. But I'll tell you, there is a pretty serious demand for the offering. I know our enthusiasm for it hasn't changed.”at 12:23
Space Exploration TechnologiesBullish
SpaceX's compute buildout will be huge, about a quarter of US gigawatts, and needs roughly $400B of capital.
On the earnings call SpaceX said it is standing up 200,000 GB300s per month, versus what used to be a year's worth of work, and expects 3 gigawatts by year-end and 10 gigawatts of compute next year. Gerstner notes the forecast for total US compute stood up next year is 43 gigawatts, so SpaceX alone would be about 25% of it. Funding that buildout needs roughly $400B of capital. He reads this as proof that Musk sees the company at the brink of superintelligence, and says whoever has the most capital, compute and data wins. He calls the biggest market debate whether there is enough capital and manpower to build US data centers. He also says SpaceX's free cash flow will help fund superintelligence. As an owner of both SpaceX and Tesla, he is bullish on the scale this buys.
“He says he's gonna do that every single month. By the end of the year, three gigawatts. He says they're gonna stand up 10 gigawatts of compute next year. It's unheard of.”at 19:29
Squawk Pod (CNBC Squawk Box): Musk Floats Mega-Merger, United CEO on Airfares & Fuel 10/21/26 ·
We are entering superintelligence, and the winner will be whoever has the most capital, compute and data.
Gerstner says Musk's announced compute ramp shows Musk understands the industry is at the brink of the singularity and entering superintelligence. The winner will be whoever has the most scale in capital, compute and data. He says most people in the industry forecast that demand will accelerate, even though the future distribution of outcomes is uncertain. Demand is unconstrained today, while supply is limited by physical constraints on power and chip production.
“It makes very clear. Elon understands we're at the brink of the singularity. We're entering super intelligence, and whoever has the most scale... Capital, compute, and data wins the game of super intelligence.”at 20:22
Squawk Pod (CNBC Squawk Box): Musk Floats Mega-Merger, United CEO on Airfares & Fuel 10/21/26 ·
CoreWeaveUncertain
CoreWeave is an extraordinary company but subscale, so it pays more for capital than SpaceX would.
Gerstner says the market's recent swings in neocloud stocks reflect scale. CoreWeave is run by a great CEO and team, but it is subscale relative to SpaceX, which has better access to capital. He explains that lenders and equity investors apply a discount rate to uncertain future demand and charge a risk premium, especially when the borrower has little to collateralize beyond its GPUs. Those GPUs are easier to underwrite when they are Nvidia parts that many buyers can use, and harder when they are custom ASICs. Since CoreWeave lacks SpaceX's scale and financing access, it faces a higher cost of capital and more financing risk.
“The problem is their subscale relative to SpaceX. ... when you're investing in somebody who doesn't have a lot to collateralize other than the GPUs.”at 20:47
Squawk Pod (CNBC Squawk Box): Musk Floats Mega-Merger, United CEO on Airfares & Fuel 10/21/26 ·
TeslaBullish
A SpaceX-Tesla deal will need a premium above 30%, likely $420 to $550 a share.
Gerstner says control premiums are typically 20 to 30%. He expects the Tesla special committee to go above 30%, because Musk sits on both boards and the committee has a unique fiduciary duty. A higher premium insulates the board from lawsuits. He also says the merger premium is partly already built into the stock, so it need not be measured from today's price. He estimates the final deal price at between $420 and $550 a share. He says scale would help SpaceX borrow and raise capital more cheaply, which he sees as one main motivation for the merger. As an owner of shares in both companies, he expects a premium for Tesla holders.
“I actually think they'll air on the side of a higher than 30%... But I think you're going to see something between, you know, $420 a share and let's call it 520 or $550 a share.”at 22:11
Squawk Pod (CNBC Squawk Box): Musk Floats Mega-Merger, United CEO on Airfares & Fuel 10/21/26 ·
NVIDIABullish
Nvidia's glut bears have been wrong for two years; supply stays constrained by wafers and power.
Asked when the shortage will turn into a glut, given margins of 75% at Nvidia and 86% at Micron, Gerstner says the glut callers have been wrong for two years and missed the upside. Nvidia is up 18x over the last three years. He argues there is a natural constraint: wafer production in the US can only grow so fast, and so can bringing power on. Unlike 2000, when demand was limited, today's demand for superintelligence is unconstrained while supply is physically limited. He also says Nvidia's GPUs are easier to finance than custom ASICs because many buyers can use them as collateral. Eventually a glut could come, but he thinks it will be far less than what the bears expect.
“Remember, Nvidia, over the course of the last three years, is up 18 x. Right. You missed all of that because you were panicked two and a half years ago that there was going to be a glut. I actually think you have a natural constraint here.”at 25:00
Squawk Pod (CNBC Squawk Box): Musk Floats Mega-Merger, United CEO on Airfares & Fuel 10/21/26 ·
Starlink's odds of disrupting carriers have risen sharply, so carrier multiples should compress.
Gerstner says the market sold off AT&T after its solid earnings because it is pricing a large disruption to carriers, not current results. The market is not saying SpaceX will certainly take them out, but the probability that SpaceX can compete with them has risen sharply. He expects SpaceX to become an alternative carrier with fast diffusion and high penetration, since carriers still leave dead zones. He argues that SpaceX's engineering and free cash flow will solve the coverage problem. So carriers' discount rates must rise and their multiples must come down. He calls the carriers' pushback a classic innovator's dilemma.
“I'm not surprised that you report earnings and your stock goes down because the market doesn't really care about what your earnings are today. They're saying there's this massive disruption coming to your business.”at 26:47
Squawk Pod (CNBC Squawk Box): Musk Floats Mega-Merger, United CEO on Airfares & Fuel 10/21/26 ·
AnthropicBullish
Anthropic's IPO could value it at $1.5T to $2T, based on 10-12x next-year GAAP revenue.
Gerstner's Altimeter forecast puts Anthropic's next-year GAAP revenue at about $140-150B. He cites the company as high growth, cash-flow positive last quarter and fairly profitable per token. At an investor day in San Francisco with about 150 leading investors, he said sentiment was 10 out of 10, with expanding competitive moats and strong compute additions. Applying a 10x multiple gives about $1.5T and 12x gives about $2T, which he sees as the valuation range for the IPO.
“But if you apply to 10 times multiple to that, Andrew, you get to 1,500,000,000,000. If you apply 12 times multiple, you get closer to 2,000,000,000,000.”at 38:55
Squawk Pod (CNBC Squawk Box): Musk Floats Mega-Merger, United CEO on Airfares & Fuel 10/21/26 ·
OpenAIBullish
OpenAI will very likely go public because it needs cheap capital, at lower multiples than past mega IPOs.
Gerstner says a public listing is highly probable. He argues that the race to superintelligence is a recursive, capital-intensive battle, so OpenAI will need access to cheap public capital. He also argues it is good for America when retail shareholders can own these companies. He points out that the IPO multiples for these companies are lower than those of Facebook, Google and SpaceX at their listings, so he sees a wall of worry already built in.
“I would say, from my perspective, it's highly probable they go public.”at 39:19
Squawk Pod (CNBC Squawk Box): Musk Floats Mega-Merger, United CEO on Airfares & Fuel 10/21/26 ·
US frontier-model safety regime is the most robust in the world and is already slowing releases.
Gerstner describes three layers. Commerce reviews every frontier model before release. A White House accord on superintelligence requires internal and external audits and board oversight, similar to Sarbanes-Oxley for AI companies. A task force led by Jay Clayton covers critical industries and cyber. He says the US is the only country to sanction a model and pull it using export controls. He credits pre-release review for the slower pace of frontier model releases, and says the framework was set up in record time instead of waiting a year for Congress.
“We don't have just a two page document. We have the most robust regulatory scheme in the world.”at 40:24
Squawk Pod (CNBC Squawk Box): Musk Floats Mega-Merger, United CEO on Airfares & Fuel 10/21/26 ·
AI will reach AGI within the next year as recursive self-improvement accelerates the race.
Gerstner says the industry is in the foothills of the singularity and that things are accelerating. Frontier labs are racing to superintelligence, and once enough compute is built they enter a recursive loop that extends the leader's lead. He says Elon Musk, Larry Page and others think AGI is not as far off as most people believe, and he says AGI is coming within the next year.
“But within the next year, we're at AGI.”at 41:44
Squawk Pod (CNBC Squawk Box): Musk Floats Mega-Merger, United CEO on Airfares & Fuel 10/21/26 ·
Trump accounts and stock gifts will channel trillions of dollars of philanthropy into children's accounts.
Gerstner says the platform is the most disruptive direct charitable vehicle in the country. Single-page donor agreements with the Treasury let donors sponsor any zip code or state and send money straight to kids, cutting out intermediaries. He says America is the most giving country and that giving has been hard, so this should draw large bequests. He also says OpenAI and Anthropic will contribute, citing Anthropic's document about giving 5% of funding to citizen accounts.
“We're gonna see trillions of dollars of bequests and other things flow through this platform because the reality is America is the most giving country on the planet.”at 43:24
Squawk Pod (CNBC Squawk Box): Musk Floats Mega-Merger, United CEO on Airfares & Fuel 10/21/26 ·
Expects choppy AI trade but plans to start buying over coming weeks as prices become enticing.
Some chop and consolidation was expected and does not surprise him. He is getting more positive and leaning in because prices are coming in and becoming enticing again, so he may start buying over the next couple of weeks.
“Have to acknowledge that there's gonna be chop along the way... Why am I leaning in and maybe start buying over the course of the next couple weeks because I'm seeing prices come in and start to get enticing again.”at 10:29
CNBC Halftime Report: AI debt Fears Fuel Tech Volatility: The Investment Committee's Strategy 10/16/26 ·
Anthropic's revenue far beat early-year forecasts, driven by Claude Code; growth should continue.
Gerstner points to Anthropic's forecast at the start of the year of about $30B, against reports that it is now doing at least $100B, as evidence that AI revenues exploded this year. He credits the Opus 4.5 breakthrough in December with leading to Claude Code, which he says replaced software engineers and drove the revenue, and argues all of the revenue derives from model capability. He agrees with Nvidia's Jensen Huang that revenue should at least double next year, since the labs' completed models will unlock further products like Claude Code.
“Remember, the forecast for Anthropic at the start of this year, $30,000,000,000. Now a lot of the reports have them doing at least a $100,000,000,000.”at 2:40
CNBC Halftime Report: The AI Trade in Focus: Altimeter's Brad Gerstner joins "Halftime Report" live in San Francisco 10/15/26 ·
AI is accelerating via recursive loops; Claude Code-style breakthroughs will spread across industries in 2027.
Gerstner argues AI is accelerating because it has entered a recursive loop in which machines build most of the code (he cites 80-90%) that drives the next product release, which some call the singularity. He expects this to show up in 2027, after the Opus 4.5 breakthrough that led to Claude Code. He argues that similar breakthroughs are coming for cyber, knowledge work, bio and materials science, creating Claude Code moments across many industries.
“I think it's accelerating. It's accelerating because we're entering a recursive loop where the machine is building more of what comes next.”at 13:16
CNBC Halftime Report: The AI Trade in Focus: Altimeter's Brad Gerstner joins "Halftime Report" live in San Francisco 10/15/26 ·
AnthropicBullish
Anthropic's IPO could draw over $500B of demand; the stock would do fine.
Gerstner says the mood in the room at Anthropic's investor day was very bullish, and the information he has gathered makes him more bullish on Anthropic, its IPO and the road ahead. He benchmarks the SpaceX IPO, which he says had about $100B offered and roughly $350B of demand, and expects Anthropic's offering, targeted before Thanksgiving pending the S-1 flip, to draw over $500B. He argues the IPO price matters little to holders over a year or two, and that retail investors, long-only and mutual funds and sovereigns should get the chance to buy.
“I suspect here you would have over $500,000,000,000 of demand for this offering. So I think the stock would do fine.”at 17:36
CNBC Halftime Report: The AI Trade in Focus: Altimeter's Brad Gerstner joins "Halftime Report" live in San Francisco 10/15/26 ·
Chinese distillation of US frontier models is a national security threat; Anthropic is working to stop it.
Gerstner says there has been significant discussion in Washington about labs cutting back on distillation, mostly by Chinese models running millions and billions of dollars of queries against US models. Those queries generate revenue, but they let competitors catch up, which he calls a national security threat. He says Anthropic is devoting significant compute and intellectual resources to stopping this, led by Dario Amodei. He estimates that if this shifts five to 7% of revenue, billions would move from Q2 into Q3, which could change how the quarter's numbers look.
“There's been a lot of conversation in Washington and other places about the need for the labs themselves to cut back on the distillation. This is Chinese models mostly running lots, millions and billions of dollars of queries against these models, which generate revenue.”at 18:37
CNBC Halftime Report: The AI Trade in Focus: Altimeter's Brad Gerstner joins "Halftime Report" live in San Francisco 10/15/26 ·
Frontier labs' lead is widening, not shrinking, because they will add the most compute next year.
Gerstner says the consensus view that the frontier lead is shrinking is wrong. He calls this the most competitive market in technology, with many open-source and new entrants posting results. He argues that the labs have models finished internally that are generations ahead of released open-source models, which is why they can claim benchmark results. He points to recursive loops of intelligence accelerating and notes that intelligence is the log of compute, and that the frontier labs are adding the most compute next year.
“And I am on the exact opposite side of that. When you think about recursive loops of intelligence, they increase at an accelerating rate. And intelligence is the log of compute. These guys are adding the most compute next year.”at 20:16
CNBC Halftime Report: The AI Trade in Focus: Altimeter's Brad Gerstner joins "Halftime Report" live in San Francisco 10/15/26 ·
Labs will cut lower-tier model prices to match open source but keep premium frontier pricing.
Gerstner argues each model generation gets much more capable on the same compute, so labs can either keep the margin or share it back to compete. He cites AWS, which cut prices about 100 times over ten years to expand its moat and benefit both customers and itself. He expects closed-source lower-tier models such as Haiku and Sonnet to become as cheap as or cheaper than open-source alternatives, while frontier intelligence should still command high prices.
“What I would say is look at their lower level of models, Haiku and Sonnet. They're going to have closed source models that are cheap... As cheap as or cheaper than the open source models.”at 22:22
CNBC Halftime Report: The AI Trade in Focus: Altimeter's Brad Gerstner joins "Halftime Report" live in San Francisco 10/15/26 ·
Enterprise AI adoption & AI software
Every business will use AI; about half of usage will be frontier models, 20-30% open source.
Gerstner says the AI market is massive and just beginning. He argues every company, large or small, will use AI, with about 20-30% of usage going to open-source models and about 50% over time going to frontier models used by large companies such as Lilly and Bristol. He draws on twenty-five years of market experience to call this the largest opportunity he has seen.
“There is not a single person or a single company, small, medium, or large, that's not going to use AI.”at 24:29
CNBC Halftime Report: The AI Trade in Focus: Altimeter's Brad Gerstner joins "Halftime Report" live in San Francisco 10/15/26 ·
CoreWeaveBullish
CoreWeave's debt is mostly cheap and GPU-secured; buying equity and debt looks attractive now.
Gerstner says most of the reporting on CoreWeave's debt is misinformation. The 15% figure is the public mark on about 20% of its debt, which it still pays about 9% on, while roughly 80% of what it borrows is secured by GPUs and is rock solid and relatively cheap. He calls the management team one of the best in the business and praises its partnerships with Nvidia. He sees the selloff as a washout, with weak hands from venture-backed and pulled IPOs being shaken out, and says he would buy the stock today. He does not own it currently.
“I think there's probably opportunity today in CoreWeave in both the equity and the debt.”at 26:29
CNBC Halftime Report: The AI Trade in Focus: Altimeter's Brad Gerstner joins "Halftime Report" live in San Francisco 10/15/26 ·
AI stock churn and consolidation avoid bubbles and separate winners from losers via multiples.
Gerstner says the churn and consolidation in AI stocks is healthy and avoids bubbles. He cites CoreWeave, which went from an IPO near $30 to about $180 when the market gave it an AI multiple, and is now around $60 as investors treat it as a GPU rental business that should not carry the multiple of Anthropic or SpaceX. He sees that repricing as the market separating winners from losers and applying appropriate multiples.
“And that is the type of market behavior that we want. It's separating winners from losers. It's applying appropriate multiples.”at 28:04
CNBC Halftime Report: The AI Trade in Focus: Altimeter's Brad Gerstner joins "Halftime Report" live in San Francisco 10/15/26 ·
US already has layered AI oversight; calls for an FDA-style AI regulator misread the situation.
Gerstner argues the US is not ignoring AI safety. A Commerce Department-led framework reviews every frontier model before release, and the US has pulled a frontier model from production earlier this year. The White House superintelligence accord is a Sarbanes-Oxley-style approach with audits and board oversight, though nonbinding, and a third body led by Jay Clayton will oversee critical industries and cyber. He says senators who call for an FDA for AI ignore this work. He accepts that some of this may be codified into law later, and that the framework will evolve, but says the right steps were taken quickly.
“We have a frontier model review framework, that is led by the commerce department. Every frontier model is gonna get reviewed before it gets released.”at 32:20
CNBC Halftime Report: The AI Trade in Focus: Altimeter's Brad Gerstner joins "Halftime Report" live in San Francisco 10/15/26 ·
A central FDA-style regulator for AI would set the US back; no group has the right answer.
Gerstner says AI governance needs to be thoughtful and developed collectively, but no single group of humans has the platonic truth on how to set it up. He rejects the idea of an FDA-style regulator that issues the rules for running AI, arguing it would slow progress. He suggests the most capable people to monitor AI are the models' own cyber capabilities.
“Do I think there's an FDA for AI that can give us the 10 commandments on how we run? No. I think that'll set us back.”at 35:29
CNBC Halftime Report: The AI Trade in Focus: Altimeter's Brad Gerstner joins "Halftime Report" live in San Francisco 10/15/26 ·
If China gets to frontier AI first, the US is not safe, so the US must stay ahead.
Gerstner argues that there is no safety off the frontier. If the US ends up in second place and China reaches frontier AI first, he says the US cannot be safe. He notes that Xi and Trump have met twice this year and that AI was discussed, and says he is hopeful about coordination. He still treats the race as central and does not treat the coordination as a substitute for US leadership.
“There is no safety off the frontier. If we end up in second place, if China gets there first, I can guarantee we're not safe.”at 35:56
CNBC Halftime Report: The AI Trade in Focus: Altimeter's Brad Gerstner joins "Halftime Report" live in San Francisco 10/15/26 ·
SnowflakeBullish
Data platforms differ from application software, so Snowflake deserves to be owned despite the software selloff.
Gerstner says he has kept Snowflake as one of his biggest personal holdings even though his hedge fund no longer owns it. He argues that IGV treats all software as equal when it is not, and that the data platform is structurally different from application software. He cites the stock being up about 50% this year as evidence that the thesis has worked so far.
“It's one of my biggest personal holdings. So great job this year. Stock's up 50%. ... part of the reason that I continued to own it in a big way was this difference between you talk about IGV as though all software is created equal.”at 42:34
CNBC Halftime Report: The AI Trade in Focus: Altimeter's Brad Gerstner joins "Halftime Report" live in San Francisco 10/15/26 ·
AnthropicBullish
Anthropic will be an enduring great company, and its IPO could come within weeks.
Gerstner calls an Anthropic IPO this year his non-consensus trade, despite negative chatter. He argues that the Google and Amazon of this era have become clearer to him, and that Anthropic will be one of the enduring great companies in the country over the next decade. Asked for timing, he says he thinks the IPO is weeks away rather than months.
“If we have an Anthropic IPO this year, right, that's my trade. It's nonconsensus... I think Anthropic will be one of the enduring and great companies in the country... I think we're talking, you know, weeks, not not months.”at 50:12
CNBC Halftime Report: The AI Trade in Focus: Altimeter's Brad Gerstner joins "Halftime Report" live in San Francisco 10/15/26 ·
AI infrastructure & data centers
Compute build-out is double-exponential, not linear; OpenAI and Anthropic may 3-5x compute next year.
Brad says analysts model companies and the economy with linear deceleration and do poorly at modelling acceleration. He sees a double exponential, with large amounts of new compute arriving. He notes OpenAI and Anthropic each have about two to three gigawatts today, which he calls small relative to what is coming. If both triple or quintuple compute next year, he argues the resulting breakthroughs, such as solving Navier-Stokes, would make today's intelligence look like the worst we will have.
“the way the economy generally works, or people model the economy, or they model a company, is linear deceleration. They're not very good at modeling acceleration. But here we have this double exponential.”at 15:09
BG2Pod: Jensen Huang | Does AI Revenue Justify Capex, '27 AI Bottlenecks, Singularity, Open vs Closed & more ·
AI infrastructure & data centers
Lab revenues must double annually, to $400B then $800B, to pay back trillions in AI capex.
He adds up the two frontier labs and several inference labs and gets about $200B of revenue this year. For the trillions of dollars of CapEx to pay back, that revenue must double next year to about $400B and double again to about $800B. He cites the roughly 10x jump in Anthropic's revenue this year as proof the labs can grow this fast. He frames this as the central test of whether the build-out pays off.
“We need to see that double next year to $400,000,000,000 We need to see it double the year after that to $800,000,000,000 in order to get in the zip code of the payback on trillions of dollars of CapEx.”at 1:01:14
BG2Pod: Jensen Huang | Does AI Revenue Justify Capex, '27 AI Bottlenecks, Singularity, Open vs Closed & more ·
NVIDIABullish
Nvidia trades at only 13-14x earnings despite consensus estimates rising 100-200% in a year.
Gerstner notes that consensus for Nvidia has risen 100-200% in a year, from a forecast of flat or shrinking revenue in 2027 to 74% growth, and that free cash flow forecasts have doubled. Yet the stock is only up about 20% and trades at 13-14x earnings. He says such a large gap between estimate revisions and the share price is unusual in his 20 years of investing, and asks what the market is pricing in. The implication is that the market is underestimating Nvidia's earnings durability.
“The consensus forecast has shifted up by a 100 to 200% for this company. Mhmm. Extraordinary... And yet the stock's only up 20%... You're now trading at 13 to 14 times earnings.”at 1:18:36
BG2Pod: Jensen Huang | Does AI Revenue Justify Capex, '27 AI Bottlenecks, Singularity, Open vs Closed & more ·
1 older view from Brad Gerstner is on the episode pages, and all of them are a question away in your AI with Chorus.
Shows
- CNBC Halftime Report · 2 episodes
- CNBC Closing Bell · 1 episode
- Squawk Pod (CNBC Squawk Box) · 1 episode
- BG2Pod · 1 episode
Ask your AI what Brad Gerstner thinks
Chorus gives Claude, ChatGPT and other assistants every view Brad Gerstner has shared on the shows it covers, next to everyone else's. Ask where they stand, what changed, and who disagrees.
Chorus reports what speakers said. It isn't investment advice.