Anthropic's $2T IPO, Zuck's AI Manifesto, Nvidia's $500B AI Bet, Grok's Comeback
David Sacks, Gavin Baker
14 Aug 20265 min read1h 15m
TL;DR
Anthropic is targeting a $2 trillion IPO with annualized revenue between $100-$120B — a 10x year-over-year growth rate sustained for three years — while insiders suggest it could realistically trade to $3-4T. Gavin Baker and David Sacks debate whether Anthropic can 10x again to $1T ARR, with the key constraints being energy infrastructure and compute supply rather than demand. Meanwhile, Zuckerberg's 6,500-word AI manifesto makes the case for decentralized, open-source AI, positioning Meta as the ideological counter to Anthropic's centralized safety-first approach.
Key Moments
Gavin Baker
“What's wild to me is that probably on the margin, Anthropic is losing share to OpenAI, to open source, and to Grock, and they're still growing so fast.”
Baker making the point that the AI market is expanding so rapidly that Anthropic can lose market share percentage-wise and still post extraordinary absolute growth numbers.
“What they believe is that future is dangerous and only under their enlightened control can humanity be protected.”
Sacks explaining the effective altruist worldview inside Anthropic — that they race to build AGI precisely because they think only they can do it safely.
All-In is a top-ranked podcast hosted by a rotating group of tech investors and entrepreneurs including Jason Calacanis, David Sacks, Chamath Palihapitiya, and David Friedberg. The show covers tech, business, and politics with a venture capital lens. This episode features David Sacks and Gavin Baker discussing the biggest AI and tech stories of the week.
Takeaways
1
Open-source AI is 90% cheaper — enterprises will notice GLM2 and other open-source models now undercut Claude Opus pricing by roughly 90%, and corporate America has a 20-year track record of eventually choosing open source over proprietary tech to avoid vendor lock-in. Founders and early adopters are already switching, suggesting Anthropic's revenue growth rate will face meaningful headwinds in 2027.
2
Anthropic insiders believe $1T ARR is possible Multiple sources close to Anthropic told Gavin Baker the company believed it could 10x from ~$60B ARR to $600B within a year, and current numbers above $80B haven't slowed. If the same 10x trajectory holds, $1T ARR by end of 2027 is mathematically plausible — which would make Anthropic a peer to Microsoft Office or Google Search.
3
Energy infrastructure is the real AI bottleneck Demand for tokens is growing exponentially and supply of compute is constrained not by chips but by physical energy — turbines, gas generators, and grid capacity. Operators are already pulling jet engines off decommissioned planes to repurpose as data center turbines, and Elon Musk reportedly bought a turbine company personally to accelerate the buildout.
4
Zuckerberg reframes AI safety as a power-concentration problem In a 6,500-word manifesto, Zuckerberg argues that the EA/Anthropic worldview — that AI is so dangerous only a few enlightened actors should control it — is itself the danger, historically leading to totalitarian outcomes. His counter-position: open-source, decentralized AI is the only safe path, and Meta is its champion.
5
OpenAI's growth is reaccelerating after coding pivot Sacks reports hearing that OpenAI's growth rate has jumped from 3-4x annually to over 20% month-over-month in the past two months — equivalent to a 10x annualized rate — after pivoting hard to coding with GPT 5.6. This suggests the coding-first strategy Anthropic pioneered with Claude is now being successfully replicated at scale.
6
Anthropic's $2T IPO price may be a banker sandbag Baker argues the $2T figure leaked to the FT likely comes from banks that lost the lead-left mandate, trying to make the winning bank look bad. His own view is Anthropic could realistically trade to $3-4T post-IPO, and pricing at $2T would be a deliberate under-price to absorb lockup volatility and keep employees focused.