Brad Gerstner | All In Summary

1. Key Themes

The AGI Threshold Has Been Crossed — And Revenue Proves It

The hosts debated whether Anthropic's Mythos model represents a genuine step-change in AI capability, with Brad Gerstner arguing this is the beginning of AGI-class models. More importantly, the revenue data validates the thesis that AI has moved from experimental to essential.

"They represent the beginning of what I would call AGI models. These are models with massive step function improvements and intelligence. And they're just too smart to be released immediately." — Brad Gerstner 00:07:32

"We are seeing...the largest revenue explosion in the history of technology." — Brad Gerstner 00:44:39

Anthropic went from $1B ARR at end of 2024, to $4B in mid-2025, $9B at end of 2025, and $30B in April 2026 — all driven by enterprise demand, not marketing.

"It was companies demanding the product. They're getting throttled on the product. Why? Because it's so good. It makes them better at their business." — Brad Gerstner 00:46:07

The Enterprise Tech Debt Problem Is Vastly Underappreciated

Chamath made a strong, contrarian case that AI coding is still in its early infancy when it comes to real enterprise use. The 50 years of accumulated tech debt in production systems is a massive, largely unsolved problem.

"The ugly truth is I don't care what model you have, but the long horizon ability for any of these models to actually build enterprise-grade software is still shit... When I call on our customers, half a trillion dollar banks, hundred billion dollar insurance companies, none of these guys are like, wow, it just works out of the box." — Chamath Palihapitiya 00:34:07

"One of our customers... they have to go and get 60-year-old pensioners to come into the office to interpret code. No, I'm not joking. This is a hundred billion dollar a year revenue company." — Chamath Palihapitiya 00:35:30

AI Value Capture Is Migrating Up the Stack — Application Layer Is the Next Battleground

The group tracked AI value capture moving from chips (NVIDIA) to hyperscalers to model companies, and now debating whether it would accrue at the application layer.

"Remember where we started? It was really just the chip layer of the stack where all the value capture was... Now we're seeing at the model layer, it looks like Anthropic and OpenAI are all going to be huge beneficiaries. I think the next question is at the application layer of the stack." — David Sacks 01:05:30

"Is it all going to be at the model layer? Or will you see an explosion of value at the application layer? I'm hoping it'll be at all layers of the stack." — David Sacks 01:05:59

2. Contrarian Perspectives

Anthropic's Safety Announcements Are a Sophisticated Go-To-Market Tactic, Not Purely Genuine

Sacks laid out a documented pattern of Anthropic using fear-based announcements to market model releases, pointing to the prior "blackmail study" as a likely reverse-engineered hoax.

"Anthropic has proven that it's very good at two things. One is product releases. The second is scaring people... I went back to Grok and I just asked, hey, give me examples where Anthropic has basically used scare tactics. And it's a pattern." — David Sacks 00:10:16

"I would say the proof that it's reverse engineered is we're now a year later. There's a bunch of open source models out there that have the same level of capability... and have you seen any examples of blackmail in the wild? I don't think so." — David Sacks 00:11:16

Chamath further reinforced with historical precedent, noting GPT-2 in 2019 had the same end-of-days narrative that became a "nothing burger." 00:15:27

Anthropic May Be Approaching "Accidental Profitability" — The Market Doesn't Know This

Brad Gerstner made a largely unnoticed claim: Anthropic's gross margins are exploding because their compute costs are largely fixed, and their revenue has grown so fast they may not be able to spend money fast enough.

"They may not be able to spend this revenue fast enough, Chamath, on compute. And remember, it's only 2,500 people. Google crossed this revenue threshold when they had 120,000 people. The only thing you can really spend money on, right, is compute. And they can't stand up the compute fast enough." — Brad Gerstner 00:51:48

"I've seen rumored out there 50 to 60% [gross margins]. The trend is going well." — Brad Gerstner 00:52:32

Open Source AI Will Win the Token War and Undercut Frontier Model Pricing

Jason argued, and Chamath partially agreed, that open source (including crypto-native distributed training approaches) will take the majority of token usage and potentially undercut all frontier model economics.

"I believe open source is going to win the day on the large language models and take 90% of the token usage. And I think the entire frontier model space could be undercut by open source." — Jason Calacanis 00:31:03

"65% to 70% of their token consumption is open source model... These revenue ramps are happening while the world is already using open source." — Brad Gerstner 00:47:01

Coding AI Market Share Could Lead to an Unbreakable Flywheel — And an Antitrust Trap

Sacks made a specific and underappreciated point: Anthropic's early lead in coding (~50-60% of coding tokens) could create compounding data advantages that lock in the lead. But it also opens them to antitrust exposure if they price discriminate against third-party tools like OpenClaw.

"It's possible that if you're the early leader in coding... you have the most access to code bases. You might get the most training tokens. There is a potential flywheel there where you could see the early market leader consolidating its lead." — David Sacks 00:40:36

"I think all these companies need to behave in a very clean way... eventually the government is going to look at this market with the benefit of 2020 hindsight." — David Sacks 01:41:30

3. Companies Identified

Anthropic

AI safety and model company. Mentioned for the fastest revenue ramp in tech history, extraordinary enterprise adoption, and the Mythos model's capabilities. Brad Gerstner invested at ~$130-150B valuation, now projecting $80-100B exit revenue run rate by end of year.

"I would not be shocked if you see Anthropic exiting this year at $80 to $100 billion in revenue." — Brad Gerstner 00:48:00

OpenClaw

Open source agentic coding tool, #1 project in GitHub history. Mentioned as being systematically undermined by Anthropic through pricing changes and product cloning, and acquired talent absorbed by OpenAI.

"OpenClaw is so powerful. It's got so much momentum that not only is Anthropic trying to ankle it... I believe when Sam Altman hired Peter, it was to subvert the open source project to get Peter's next set of genius ideas inside of OpenAI." — Jason Calacanis 00:29:17

Ridges AI (Subnet 62 on Bittensor)

Distributed open source AI coding project. Mentioned as a credible, fast-moving threat to Claude Code, reaching 80% of Claude 4 capability in 45 days for $1M in TAO rewards.

"There's a project that's subnet 62. It's called Ridges AI... They spent about a million dollars in Tau, like rewards. And in 45 days, they hit 80% of what Claude 4 is." — Jason Calacanis 00:36:54

Venice

Distributed open source training and orchestration project. Briefly mentioned by Chamath as a potentially hugely disruptive alternative to the capital-intensive frontier model training model.

"Another project other than BitTensor that someone brought up to me is Venice. The concept of open source training and orchestration is a hugely disruptive idea, which is the complete orthogonal attack vector to this idea that you have to raise tens and tens of billions of dollars to train your models." — Chamath Palihapitiya 00:38:04

X (formerly Twitter)

Social media platform. Mentioned specifically for its auto-translate feature powered by Grok, which the hosts praised as transformative for cross-border information access.

"Because of Grok being really good at doing auto-translate, they've taken the pockets of the best of what's happening in Japan, what's happening in Israel, what's happening in France, and they're surfacing it auto-translated." — Jason Calacanis 01:22:08 "X is better today than it's ever been. And remember, they have 70% fewer employees than they had the day Elon walked into the building." — Brad Gerstner 01:24:25

Athena

Executive assistant staffing company (Philippines-based). Mentioned as generating 1,000+ signups after a prior episode mention and now hiring 500 new assistants as a result.

"They had 1,000 people after last week when we mentioned how much we love Athena." — Jason Calacanis 01:27:52

Palantir

Enterprise AI application company. Cited by Sacks as an example of an application-layer company already turbocharged by AI model capabilities.

"You could say that Palantir is already one of them, right? It's an application company that's been turbocharged by these model capabilities." — David Sacks 01:05:59

4. People Identified

Dario Amodei

CEO of Anthropic, former OpenAI co-founder. Praised for disciplined product focus (no multimodal, no video, no hardware) and the Mythos safety sandbox approach.

"Anthropic made choices. No multimodal, no video, no hardware, no chips, no building data centers. They said, we're just going to focus on coding and co-work. We think that is the path to AGI and ASI. They executed their butts off." — Brad Gerstner 00:59:03

Peter Steinberger

Founder of OpenClaw, described as a "renowned coder" who created the #1 open source project on GitHub. Now acquired by OpenAI. Mentioned as a central figure in the agentic AI era.

"Renowned coder who created OpenClaw, which is kind of the thing that launched this whole agent era now." — David Sacks 00:24:18

Elisha Long ("Eli")

YouTube philosopher and modern sage. Mentioned as a rising figure discovered by Marc Andreessen, with content resonating at the highest levels of Silicon Valley.

"Marc Andreessen found him. And he's like, this guy is the new guy... He's the Lisan Al-Gaib of the modern internet." — Chamath Palihapitiya 00:03:47 "The more you want something, the less you're going to get it... that detachment is really healthy for people." — Chamath Palihapitiya 00:03:04

Naftali Bennett

Former Israeli Prime Minister. Mentioned for publicly acknowledging falling Israeli popularity in the US, suggesting even Israeli political figures recognize the geopolitical cost of the current path.

"Naftali Bennett, who is a major Israeli politician who was a former prime minister, tweeted polling that showed that Israel was becoming very unpopular in the U.S. And he was expressing concern about that." — David Sacks 01:21:12

5. Operating Insights

The "Accidental Profitability" Model: Fixed Compute + Exploding Revenue = Surprise Margins

For operators building AI-native businesses: if your primary cost input is compute (largely fixed in the near term), and revenue is scaling faster than you can deploy new infrastructure, gross margins can improve dramatically and counterintuitively. This is a non-obvious structural advantage for lean AI companies.

"They had that gigawatt and a half of compute, whether they have a billion in revenue or whether they have 80 billion in revenue. So you might actually expect to see these companies — their gross margins are exploding higher." — Brad Gerstner 00:51:13

Barbell Portfolio Strategy for the AI Transition Era

Brad described Altimeter's explicit approach to navigating the AI disruption of enterprise software: go heavy on frontier AI companies and small defensible niche teams; avoid the middle.

"We've taken a barbell approach, right? We've got a lot in what we think are the most important companies that are on the frontier. And then we're betting on really small teams that we think have very defensible businesses in a world of AGI." — Brad Gerstner 01:03:52

Subscriber-Only Reply Mode as a Feed Quality Hack

Jason shared a tactical social media operating insight: restricting tweet replies to paid subscribers dramatically improves signal quality and builds a tighter high-quality community.

"I do 50-50. Sometimes I'll just let it rip and get chaos. And then other times I have 2,000 paid subscribers. I give all the money to charity, like 30 grand a year. And it's just wonderful to get to know the same 2,000 people out of my million followers." — Jason Calacanis 01:25:26

6. Overlooked Insights

Venice: The Distributed Training Play That Could Eliminate the $10B Fundraise Requirement

Chamath made a fleeting but potentially massive observation about Venice — a project focused on distributed open source pre-training and orchestration. If successful, this represents a complete structural attack on the capital moat that OpenAI and Anthropic have built. The implication is that the fundraising arms race ($10B+ rounds) could become irrelevant if distributed training reaches sufficient quality.

"The concept of open source training and orchestration is a hugely disruptive idea, which is the complete orthogonal attack vector to this idea that you have to raise tens and tens of billions of dollars to train your models. Because if the capital markets run out of 10 and 20 billion dollar checks to give people, the only solution is to be totally distributed." — Chamath Palihapitiya 00:38:04

Nobody on the panel pushed on this. If Venice or a project like it succeeds, it doesn't just threaten model companies — it threatens the entire venture capital narrative around AI infrastructure. Worth deep investigation.

The Real Cybersecurity Opportunity Is Real-Time AI Security Embedded in Coding Tools — Not Patching

Jason briefly noted something the group largely passed over: the more important security implication of Mythos is not the 100-day patch window, but rather the need to embed security scanning in real-time into the AI coding tools generating exponentially more code right now.

"The amount of code being pushed right now because of these tools is 10x, 100x in most organizations. So we need to have this type of security embedded in these new coding tools to do it in real time. That's the opportunity. There should be real-time correcting of this." — Jason Calacanis 00:22:04

No one followed up. This is a product gap and an investment thesis: a real-time AI security layer natively integrated into coding tools (Cursor, Claude Code, Codex, etc.) is a natural next category. The company that owns this layer owns critical infrastructure for every enterprise software development workflow.