Apr 10, 2026 · 1h 29m
Anthropic charts rapid growth as security fears delay its next AI model
Anthropic's $30B Ramp, Mythos Doomsday, OpenClaw Ankled, Iran War Ceasefire, Israel's Influence
This episode connects the commercial realities of the AI race with the geopolitical shifts shaping global stability and U.S. foreign policy.
- 1Anthropic is experiencing historic revenue growth while balancing safety demands that have delayed its next major model.
- 2The controversy surrounding OpenClaw highlights the friction between open AI assistants and established platform ecosystems.
- 3The geopolitical landscape is shifting rapidly following the Iran ceasefire and evolving U.S. policy dynamics with Israel.
The brief
The All-In hosts are joined by Altimeter Capital founder Brad Gerstner to dissect Anthropic's staggering revenue growth and the strategic decisions shaping the escalating AI arms race.
Anthropic's choice to delay its Mythos model due to safety concerns highlights a growing tension between rapid commercial deployment and the rigorous security protocols required for next-generation AI.
The panel also dives into the OpenClaw controversy, evaluating how personal AI assistants are shifting the consumer landscape while raising fresh questions about data access and platform rules.
Beyond technology, the discussion pivots to global geopolitics, analyzing the strategic implications of the Iran ceasefire and how Israel continues to influence United States foreign policy.
What was said on this episode
41 statements · 24 positive · 14 negative · 2 mixed · 1 neutral
Chamath says detachment from outcomes is healthy for people.
“I think that that detachment is really healthy for people.”
Listen at 3:29
Mythos can chain multiple vulnerabilities into sophisticated exploits.
“this model is able to create exploits out of three, four, sometimes five vulnerabilities”
Listen at 5:42
Brad says releasing Mythos immediately would wreak havoc on the Internet.
“the company realized it would wreak havoc”
Listen at 6:33
Brad predicts OpenAI and Google will sandbox similarly capable models.
“I think that OpenAI will end up doing the same thing. I think Google will end up doing the same thing.”
Listen at 8:25
Sacks says Anthropic uses fear as part of its product-release strategy.
“Anthropic has proven that it's very good at two things. One is product releases. The second is scaring people.”
Listen at 9:52
Sacks judges Anthropic’s cyber-risk warning as substantially legitimate.
“I actually think that this one is more on the legitimate side.”
Listen at 11:34
Sacks predicts AI-driven cybersecurity will uncover dormant bugs over the next six months.
“over, say the next six months, we're going to have this, call it one time period of catching up where AI Driven Cyber is going to be able to detect a whole range of bugs”
Listen at 12:10
Sacks recommends code owners use the next few months to find and patch dormant vulnerabilities.
“every company or IT department or CISO that is managing code bases should take this seriously and use the next few months to detect any again, like dormant bugs or vulnerabilities and roll out patches.”
Listen at 13:17
Sacks says coordinated patching can avert an AI-cyber doomsday scenario.
“If everybody does their job and reacts the right way, then I do not think it will be the doomsday scenario”
Listen at 13:33
Chamath judges Anthropic’s Mythos safety presentation mostly theatrical.
“I think it's mostly theater.”
Listen at 15:00
Chamath says sophisticated hackers can already achieve Mythos-like results with Opus.
“a very sophisticated hacker can probably do those things right now with opus.”
Listen at 15:54
Chamath estimates patching all alleged vulnerabilities would require shutting down the Internet for years.
“you'd have to shut down the Internet for about five years to patch them all.”
Listen at 16:10
Sacks predicts widespread patching will resolve the immediate AI-cyber problem.
“my guess is we're going to see a lot of patches over the next few months. I think that that will resolve the problem.”
Listen at 21:07
Chamath says Project Glasswing should include energy and aviation companies.
“If this is real, they picked the wrong companies.”
Listen at 22:32
Brad says Anthropic’s subscription pricing heavily subsidized OpenClaw power users.
“they were selling dollars for $0.10 via OpenClaw”
Listen at 27:03
Jason predicts open-source models will capture 90% of large-language-model token usage.
“These, I believe open source is going to win the day on the large language models and take 90% of the token usage.”
Listen at 30:53
Jason recommends against entrusting all personal or business knowledge to frontier models.
“I would be very concerned if I was you with putting all of your knowledge that you've earned over a lifetime into any of these large language models.”
Listen at 31:44
Chamath estimates Anthropic has 50–60% coding-token market share.
“They probably have 50 to 60% market share”
Listen at 33:37
Chamath estimates AI-enabled coding represents 5% of the broad coding market.
“AI enabled coding I think is still 5% of the broad market.”
Listen at 33:48
Chamath says current AI models cannot reliably build enterprise-grade software.
“the long horizon ability for any of these models to actually build enterprise grade software is still S H I T”
Listen at 34:12
Chamath predicts resolving global technology debt will take decades.
“It's going to take decades to do it right.”
Listen at 35:44
Jason says Ridges AI reached 80% of Claude 4’s capability within 45 days.
“in 45 days they hit 80% of what Claude 4 is”
Listen at 36:50
Chamath assigns essentially zero probability to major companies outsourcing production code to open source.
“I would put zero, literally the probability zero of any important company worth anything more than a dollar having and outsourcing their production code to an open source project.”
Listen at 37:24
Chamath predicts a successful open-source model pretraining project will emerge.
“there is going to be at some point a very successful open source project for pre training.”
Listen at 38:35
Sacks predicts AI will generate 95% of code over the next few years.
“I think it's going to 95% mean I bet any amount of money on that.”
Listen at 39:58
Sacks says coding-market leadership could reinforce itself through data and training-token advantages.
“you might get the most training tokens. There is a potential flywheel there where you can see the early market leader consolidating its lead”
Listen at 40:22
Sacks recommends that AI companies use fair pricing and avoid anticompetitive discrimination.
“engage in fair pricing”
Listen at 41:59
Brad forecasts Anthropic ending the year with $80–100 billion revenue run rate.
“I would not be shocked if you see Anthropic exiting this year at 80 to $100 billion in revenue.”
Listen at 48:18
Chamath says AI companies remain years away from demonstrating steady-state profitability metrics.
“We're never, we're, we're years from that.”
Listen at 50:56
Brad says Anthropic and OpenAI gross margins are rising rapidly as revenue scales.
“their gross margins are exploding higher”
Listen at 51:33
Brad predicts AI companies may become accidentally profitable because compute deployment lags revenue.
“You may see accidental why I call it accidental profitability.”
Listen at 51:48
Brad says AI inference costs have fallen 90% year over year.
“the inference cost is plummeting. Inference cost is down by 90% year over year.”
Listen at 54:37
Brad says OpenAI remains a strong contender despite recent setbacks.
“I think you would be seriously foolish to count out OpenAI.”
Listen at 59:40
Brad recommends buying OpenAI shares at current prices.
“I'm a buyer of the shares today.”
Listen at 1:00:57
Sacks evaluates the Iran ceasefire positively.
“I think it's terrific that we have the ceasefire.”
Listen at 1:12:25
Sacks credits President Trump with negotiating the Iran ceasefire.
“I give the President tremendous credit for negotiating the ceasefire that we've achieved so far”
Listen at 1:12:58
Chamath warns Israel could lose US alliance reliability without a quick conflict off-ramp.
“unless I help find an off ramp quickly, the risk that Israel loses America as a predictably steadfast ally could go down.”
Listen at 1:17:59
Sacks says Israel would improve US popularity by supporting the ceasefire.
“I think that they would improve their popularity if they got behind the ceasefire”
Listen at 1:21:38
Jason evaluates X’s automatic translation as an exceptionally impressive technology feature.
“it is the most impressive tech feature I've seen released in years.”
Listen at 1:21:58
Brad judges X’s current user experience better than at any previous time.
“X is better today than it's ever been.”
Listen at 1:24:26
Brad says technology companies can produce more output with fewer existing employees.
“for the existing jobs, these companies are all realizing they can do more with less.”
Listen at 1:24:57
Statements are attributed to the speaker as said on the episode and reflect their view at the time, not PodLume's. They are not advice.