PodcastAll-In PodcastGavin Baker

Anthropic's $2T IPO, Zuck's AI Manifesto, Nvidia's $500B AI Bet, Grok's Comeback

The discussion examines the economic and strategic stakes of the AI buildout, from Texas grid constraints and a potentially interconnected supply chain to the debate over centralized versus open-source systems. Anthropic is portrayed as the current model leader and a major market signal, with a potential $2 trillion IPO, while NVIDIA financing and large compute projects raise the risk of oversupply. Speakers also debate AI's effects on growth, profitability, employment, and productivity, alongside forecasts for Zuckerberg's open-source efforts and xAI's path toward frontier-model status.

Artificial IntelligenceAI InfrastructureSemiconductorsTechnology StocksEconomic GrowthSupply Chains

Preview · 23 more points in the app

Read the Full Summary

Key Takeaways

The speakers estimate that corporations and workers could spend roughly $8 trillion, or 5% to 10% of salaries, on AI tokens and productivity tools. They argue that frontier AI tokens could capture 65% to 85% of economic value even if open-source models generate most usage volume, while Anthropic and open-source tokens are already described as profitable.

The episode frames the AI policy debate as a choice between centralized approval and decentralized access: Anthropic and effective altruists are described as viewing AI as too dangerous to distribute, while Zuckerberg, Musk, and Huang view it as too dangerous to centralize. Speakers argue that banning open-source AI or slowing U.S. compute could weaken competitiveness against China.

Anthropic is presented as the current model leader, with Polymarket assigning it a 67% chance of remaining on top at year-end, but GLM 2 reportedly costs 90% less than Claude Opus. The speakers say Anthropic and OpenAI can sustain premium pricing only if they remain ahead, with the top 20% of customers potentially paying 10 times more.

Anthropic's reported 10-times year-over-year growth for three consecutive years is used to illustrate both the industry's upside and its physical limits: continuing that pace could theoretically produce $1 trillion in annual recurring revenue by the end of the following year, although compute and energy constraints may intervene. Its potential IPO is viewed as a market signal for whether AI demand and profitability justify the sector's capital expenditures.

The proposed $500 billion NVIDIA financing structure would let companies borrow to buy systems, rent out compute, and use the resulting revenue to repay debt. Speakers warn that this model could amplify the downside if capacity built for $30 to $50 per watt becomes uneconomic in a compute glut similar to the post-dot-com dark-fiber oversupply.

Data-center expansion faces a turbine-blade bottleneck because production is concentrated in two facilities in America and Europe, prompting consideration of repurposed jet engines and expanded capacity from Caterpillar, Cummins, and Siemens Energy. The speakers say 5-to-10-times inference-efficiency improvements every 12 to 18 months could partially offset the resulting energy and compute constraints if sustained.

GEO-POLITICS

Texas Energy Audit — Texas Governor Greg Abbott announced an energy audit for data-center projects seeking access to the Texas grid. Projects that provide their own energy would be exempt; the stated goal is grid protection, not a moratorium.

“Governor Abbott announced this week he's gonna do an energy audit…”—

Centralized AI Control — The discussion contrasts a centralized AI regime favored by some government-affiliated thinkers with decentralized systems whose values users can control. Speakers argue that centralized approval regimes could eliminate frontier-model leads and concentrate control among a few companies.

“I also dealt with think tank people who are part of this, like, click…”—

Frontier AI Value — The speaker argues that frontier AI tokens could capture 65% to 85% of economic value, even if open-source models generate most usage volume.

“But they needed, like, the 2,000, you know, merely, you know, top top 10,000 in the world businesses…”—

TECHNOLOGY

Anthropic Model Race — Anthropic is described as the current model leader, with Polymarket assigning it a 67% chance of remaining on top at year-end. The company is reportedly discussing a $6 billion Dekart acquisition to improve chip efficiency, while Grok and OpenAI catch up.

“Polymarket says 67% chance that at the end of the year, Anthropic will have the top model…”—

Data-Center Power — Data-center power infrastructure faces a turbine-blade bottleneck because production is concentrated in two facilities in the United States and Europe. Older jet engines can be repurposed as onsite turbines, while Caterpillar, Cummins, and Siemens Energy expand capacity.

“Everybody talks about, like, the ultimate constraint are these turbine blades…”—

Open-Source Pressure — GLM 2 is described as costing 90% less than Claude Opus, increasing competitive pressure on Anthropic's pricing and growth. Corporations and startups are drawn to open source because it reduces dependence on large providers.

“You just look at the pricing here of GLM 2 versus Claud Opus, and, you know, it is 90% cheaper to use these products. I'm using them…”—

ECONOMY

AI Growth Effects — AI is described as currently accelerating growth, with more software-coding openings than a year earlier, although young workers may be experiencing negative effects. Broader growth effects may appear in national accounts within 18 months.

“Thus far, it really does look like it's accelerating growth. There are, you know, there are more job openings for software coders today than there were a year ago…”—

AI Access — The speakers argue that AI is a broadly needed consumer and business technology, rather than something analogous to a nuclear weapon. They also argue that banning open-source AI or centralizing access could weaken U.S. competitiveness against China.

“Every consumer, every business needs AI. It's fundamentally different. This is a consumer technology. Yes. It's a con it's a consumer technology before it's a military technology.”—

MARKETS

GPU Financing — NVIDIA and major financial firms are reportedly developing a $500 billion financing structure. Companies would borrow to buy NVIDIA systems, rent out compute, and use the resulting income to repay debt.

“NVIDIA is partnering with Goldman, BlackRock, and others to raise 500,000,000,000 in AI compute. Jensen laid out his vision in an X article, as we mentioned.”—

Compute Oversupply — The principal downside risk identified is a glut of compute analogous to the dark-fiber oversupply after the dot-com crash. Capacity built for $30 to $50 per watt spot prices could become uneconomic if supply exceeds demand.

“The biggest risk is to me is not on the demand side. The biggest risk is that you get a glut of compute and you get an overbuild.”—

NVIDIA Financing Role — NVIDIA could become a capital-light cloud and financing intermediary by sharing compute revenues or providing seller financing for large expansions, such as xAI's proposed 6-to-8-gigawatt buildout. That project is estimated to require $300 billion to $400 billion.

“Elon wants to add somewhere around 6 to 8 gigawatts next year. We know that that would cost 3 to $400,000,000,000 of CapEx.”—

FORECASTS

Anthropic IPO — Anthropic is reportedly targeting a $2 trillion IPO, with an 80% Polymarket probability of going public this year. Speakers are skeptical that the rumored valuation is definitive, but say pricing near $2 trillion would signal a successful roadshow and strong demand.

“If they price it at 2, it means, like, the testing the waters went well, the roadshow went well, and they've got a lot of confidence about where it's gonna…”—

AI Productivity — The speaker argues that workers are more likely to lose jobs to people using AI than directly to AI itself. They claim adoption can multiply individual output by five to ten times while creating new business opportunities and jobs.

“You're not gonna lose your job to AI. You're gonna lose it to somebody with AI. And that's the game on the field right now…”—

Executive X Risk — The speaker argues that public figures and company CEOs without an authentic personal voice on X face reputational and accountability risks. Corporate PR may not protect them if leadership changes or they face congressional scrutiny.

“If you are a public figure, if you are the CEO of a company, and you do not have your own voice on x, you are at risk…”—

Gavin Baker joins the show!

The hosts and Gavin Baker assess Anthropic’s potential $2 trillion IPO, extraordinary revenue growth, competitive position, and whether compute, energy, and demand can sustain expansion.

Anthropic IPO report: $2T valuation, $100B+ run rate, October listing

The panel largely endorses Zuckerberg’s decentralized, open-source AI vision, contrasting individual empowerment and competition with Anthropic’s preferred centralized safety regime.

Zuck's AI manifesto: What it means for Meta and frontier AI

The hosts announce All-In Summit speakers, including Steve Hilton, Spencer Pratt, Abigail Shrier, Gavin Baker, and Boom Supersonic’s Blake Scholl, alongside new festival-style programming.

All-In Summit Speaker Announcements!

NVIDIA’s financing plan would make GPU infrastructure financeable through major banks and private equity, while the panel debates residual guarantees, oversupply risk, and circular-financing concerns.

Nvidia's $500B financing plan, how the AI market could fall apart

The panel argues Amazon’s subcontracted-driver model shifts costs and risk onto workers and taxpayers, while opponents warn that mandated employment could sharply raise delivery costs and reduce flexibility.

NJ and Mamdani take on Amazon over subcontracted drivers

Grok 4.6 and Grok Bot are portrayed as evidence that xAI is approaching frontier capability, aided by Cursor talent, SpaceX management, and Elon Musk’s high-upside compute strategy.

Grok 4.6 launch: SpaceX's high-ceiling, high-floor AI strategy

Workday’s possible Silver Lake acquisition signals renewed private-equity interest in software, with open-source AI potentially improving the economics of mature software businesses.

Workday in talks to be acquired by Silver Lake for ~$43B

The episode closes with Gavin’s rising bestie ranking, invitations to future gatherings, and a brief farewell after discussing the week’s major AI and market developments.

Related Episodes

Important Disclaimers

Not Investment Advice: For education only — not financial, tax or legal advice.

AI-Generated: Made by AI from source transcripts and may contain errors.

No Guarantees: Past performance does not predict future results; investing carries risk.

Our Commentary · Request Removal