All-In

More Trillion Dollar IPOs, Anthropic $3T, Zuck's Price War, China Ends Open Source?, Trump Accounts

Jason Calacanis, Chamath Palihapitiya, Brad Gerstner
11 Jul 2026 5 min read 1h 30m

Anthropic and OpenAI are both on track to IPO within 6-9 months, with Anthropic rumored at $3T and potentially $100B+ in revenue by year-end. Brad Gerstner says Altimeter would be a buyer at scale in both IPOs, while Chamath warns that enterprise AI token costs are doubling every 45 days with only ~2% measurable ROI — suggesting a reckoning is coming for corporate AI spend.

Chamath Palihapitiya
“He said, "Right now, our token costs are doubling every 45 days." And I was like, "Gh." And he said, "Yeah." And I said, 'Well, what is the downstream productivity? And he said, maybe 5% max.”
Chamath recounting a conversation with his CTO about AI token spend at his company 8090, raising concerns about enterprise ROI.
▶ 5:16
Chamath Palihapitiya
“the actual ROI was somewhere between zero and 2%.”
Chamath describing what Claude (Fable 5) told him when he asked about actual EPS lift from AI for the S&P 493, excluding Nvidia's chip revenues.
▶ 15:46
Brad Gerstner
“If these guys end a year over a 100red billion, I think that they're on a revenue trajectory that they could 3 to 5x again next year. We've never seen anything like this. Never.”
Brad Gerstner making the case for Anthropic's extraordinary revenue trajectory and why the IPO would be a blockbuster.
▶ 22:05
Chamath Palihapitiya
“once I got down to 95% cheaper I started setting my agents instead of doing daily runs to doing hourly runs then I took my agents from doing one task and I broke them up into three agents and have them doing three different things on the hour. And when you start doing hourly tasks and then you wake up in the morning and like 14 jobs have been done, you're like, "Wait a second. This is completely different."”
Chamath describing his personal experience after getting cheap API access via a Bittensor subnet, illustrating Jevons paradox in AI token consumption.
▶ 26:35
Brad Gerstner
“despite all of those arguments, the facts in the field are just the opposite. The share of economic value, right? There's this quote, there's this tweet this week from Jesse Zang that we ought to pull up here. You know, the economic value, the share of wallet is actually increasing to the Frontier Labs while the share of tokens, these commodity tokens is obviously going up, you know, to the other guys.”
Brad pushing back on the DeepSeek-era thesis that open-source would kill frontier model revenues, citing 18 months of data.
▶ 28:13
All-In is a weekly podcast hosted by Jason Calacanis, Chamath Palihapitiya, David Friedberg, and David Sacks — four friends who discuss tech, business, and politics with unfiltered opinions. The show is known for its candid debates among insiders with deep investing and operating experience. In this episode, Freeberg is absent and Brad Gerstner (Altimeter Capital) joins as a guest host.
1
Enterprise AI token costs doubling every 45 days Chamath's CTO told him their token spend doubles every 45 days while downstream productivity gains are capped around 5%. When he asked Claude to quantify AI's actual EPS lift for the S&P 493 (excluding Nvidia), the answer was 0-2% — mostly explained by pricing power and buybacks, not genuine AI productivity.
2
95% token cost drop triggers Jevons paradox immediately Chamath obtained cheap API keys via a Bittensor subnet running GLM 5.2 and watched his behavior change instantly — daily agent runs became hourly, single tasks split into three parallel agents. This illustrates that dramatically lower inference costs don't reduce spend; they explode usage and could ultimately benefit frontier labs through volume.
3
Uber deploying 'agentic pods' to find operational ROI Uber CTO Pine shared that 99% of engineers use AI tools and 70%+ of pull requests are agent-attributed — but the more interesting move is embedding engineers into non-technical departments (legal, HR, procurement, operations) to build 200 agentic skills and find cost-takeout ROI beyond just developer productivity.