Episode summary
Open Source Wins, AGI Is Here, and Scorsese’s AI Toolkit with CEOs of Cerebras & Black Forest Labs
All-In Podcast
· · 1 hr 4 min · Hosts: David Sacks, Jason Calacanis, Chamath Palihapitiya, Ben Horowitz, David Friedberg, Brad Gerstner · Guests: Andrew Feldman, Robin Rombach
Summary
The conversation centers on the AI buildout as a physical infrastructure race, with the guest arguing that today’s demand is unlike anything seen in modern technology. He says data centers are now being built across the US, Canada, the Nordics, Europe, the Middle East, and Central Asia, and that hyperscalers are placing orders so far ahead that Cerebras already has a $25 billion backlog. The host frames this as a mobilization on the scale of the war effort, and the guest agrees that the market is not speculative so much as booked demand.
From there, the discussion turns to inference, reasoning, and what the guest sees as a Moore’s-law-style trajectory for Cerebras. He argues that reasoning models consume huge numbers of internal tokens, making fast compute essential, and says Cerebras has already broken the old doubling curve and expects to stay well above 2x over the next 18 months. The host pushes on AGI, and both suggest that by older definitions it is already here, even if it is not fully deployed.
The second half shifts to Robin Rombach and Black Forest Labs, where the focus is open source image and video generation, multimodal models, and the move toward world models and robotics. Rombach describes using generative models as a medium for human creativity, including work with Martin Scorsese, and says the company is seeing demand from film, launch videos, and IP owners who want controlled, customizable generation. He emphasizes that open source, sovereignty, and human-in-the-loop workflows will be central as the technology gets faster and more interactive.
Key takeaways
- The guest says Cerebras has a $25 billion backlog because buyers like OpenAI, Anthropic, Google, Microsoft, and AWS are ordering capacity far in advance.
- The guest argues that reasoning is computationally intensive and that fast inference hardware is essential because internal token usage drives better answers.
- The host and guest both suggest AGI is already effectively here by older benchmarks, even if it is not fully deployed in products.
- The guest says open source is the right layer for many ordinary business tasks, while frontier models will handle the hardest problems.
- Robin Rombach says Black Forest Labs is building multimodal models that move from images to video, audio, and eventually action prediction for robotics.
Notable quotes
“We are in the race for super intelligence and uh Andrew Feldman is back uh and obviously CEO and founder of uh Cerebras”
“we have a $25 billion backlog”
“this reasoning consumes a huge amount of tokens internally”
“AGI I think I suspect you'll agree with me that we've hit it. we just haven't exactly deployed it fully”
“open source is kind of having a moment right now”
How the episode opens
We are in the race for super intelligence and uh Andrew Feldman is back uh and obviously CEO and founder of uh Cerebras doing inference chips pioneered the space had a successful IPO. We've talked about this a couple of times. We got to see each other in January at Davos. IPO happens. Uh the boys and I got to sit with you recently.
That was fun — At liquidity. — That was really that was really fun. — Had a great discussion with the boys. I wanted to deep dive with you about a couple of topics. The first one is the buildout of AI. We've never seen a build out like this since, you know, the Great Wall of China,
Right? Who knows? — The pyramids. I mean, it feels like the amount of capital, time, and intelligent people on the planet dedicating themselves to the buildout of something. Um I I can't think of anything in our lifetimes with per perhaps uh you know before our lifetimes the war effort
Right — This is a mobilization and a scale that we read about we hear about but you're actually doing it you have customers who are building data centers and you're a key piece of that
All app started with an $8 domain and no VC funding and became one of the largest ad platforms in the world now that same engine powers Appleven ads for e-commerce. Your ads run inside mobile games, reaching over a billion people with full screen, distraction-free attention. The platform finds buyers and optimizes for profit.
You set the target, it does the rest. One cookware brand went from 4 million to $16 million, turned profitable, and is on pace for 80 million this year. Visit apploven.com/allin to launch your first campaign today.
From the episode’s captions: roughly the first two minutes, lightly cleaned. Keep watching on YouTube →
Topics
- data center buildout
- open source ai
- agi progress
- generative video
- ai in hollywood
Companies and tickers mentioned
- Cerebras
- Black Forest Labs
Watch the episode
Summary, takeaways and quote selection by Octagon. Quotes are short excerpts; the full conversation is in the episode.
More from All-In Podcast
-
Jensen OUT as Nvidia CEO: "Spy" Scandal, Export Restrictions, Market Impact -
Freedom vs. Safety: Axon CEO Rick Smith on the Future of Policing -
Midterm Blood Bath, $4T AI Sell Off, Data Center Freeze, What’s Next for Trump -
All-In Podcast REACTS to Democrat Sweep! -
Anthropic S-1 Shocker, Greatest Business Ever? Google’s $500B Wipeout, Midterm Predictions -
WHOOP CEO, Will Ahmed: 50B Hours of Health Data, a 4 Million Member Data Moat, & the FDA Fight
All-In Podcast, this episode and its artwork belong to All-In Podcast, LLC. Octagon is not affiliated with the show; this page is an independent summary.
