Speaker
Andrew Feldman
Appearances over time
2 episodes
Episodes
2
Podcasts
Quotes & moments
Feldman argues that by any definition of AGI used 10, 20, or 50 years ago — including the Turing test — AI has already blown past it.
Cerebras has a $25 billion backlog of orders, reflecting demand that far outstrips the industry's ability to build and fill data centers.
Individual AI data centers under construction will draw more power than mid-sized cities, with aggregate demand set to exceed the previous 50 years of Earth's energy use.
While GPUs are constrained by Moore's Law (doubling every 18 months), Cerebras' novel chip architecture is on track to exceed 2x performance gains in the same period.
Feldman argues AI gives humanity a real shot that neither our children nor anyone they know will die of cancer — the defining pro on the ledger of AI's costs and benefits.
Modern reasoning models consume enormous numbers of tokens internally during inference — making fast inference hardware like Cerebras chips disproportionately valuable.
Cerebras chips can run inference 15 times faster than competing hardware, meaning a 24-hour run on Cerebras could yield weeks or months worth of AI thinking.
By resolving critical memory bottlenecks, Cerebras achieves massive speed gains on production-grade machine learning models.
Feldman referenced Thomas Kuhn's insight that scientific paradigm shifts only happen when incumbent thinkers die — a cycle AI is now compressing by accelerating learning across the equivalent of thousands of generations.
Going public involves massive administrative overhead with negligible direct impact on core operations. High-fives and celebration are great, but the day after the IPO, your engineering and supply chain projects face the exact same realities.
Cerebras spent years overcoming CFIUS concerns and geopolitical hurdles. Everything was incredibly difficult until it suddenly became easy, with the pricing range raised twice before hitting a massive public valuation.
An IPO is more than just a financial milestone; it represents a major emotional landmark for long-term employees. For many engineers and immigrant families, the public listing provides a powerful form of external validation.
For companies serving critical industries like agriculture and defense, an IPO is a maturing event. Global governments and sovereign customers need to know their primary space data suppliers aren't going to vanish overnight.
Planet Labs operates a massive constellation of 200 satellites to capture daily global updates. By tracking every change on Earth in real time, they empower farmers, civil governments, and intelligence networks with unprecedented insights.
By leveraging sun-synchronous orbits, space-based data centers can access five times more solar energy than terrestrial systems without needing batteries. With launch costs dropping toward $200 per kilogram, orbital compute is becoming financially viable.
Compute historically focused almost entirely on numerical calculations. AI represents a foundational shift, enabling silicon to natively decode, synthesize, and leverage language and images for the first time in computing history.
To beat NVIDIA, you cannot build a standard GPU. Cerebras took a contrarian route, manufacturing a dinner-plate-sized wafer chip that places memory directly next to the compute engines, eliminating the critical data transfer bottleneck.
Historically, the absolute scale of wealth generation occurs in the public markets, not the private phase. Long-term institutional investors and VCs capture far greater gains by holding onto high-growth stocks post-IPO.
Large language models are blind to physical reality because they are trained solely on the internet. By feeding physical, real-time planetary sensor data directly into space-based AI networks, we will build a highly actionable 'planetary intelligence' system.
Every chip before Cerebras followed Moore's Law — doubling performance every 18 months. Cerebras broke that curve with a fundamentally new architecture, and expects to far exceed 2x gains in the next 18 months. New architectures have room to optimize that mature 20-year-old GPU designs simply can't access.
A Bitcoin movie starring Gal Gadot was filmed entirely on a sound stage, with all scenery generated by AI in post. The result: a $30M production that would have cost $150M with traditional set builds. It never would have been greenlit at $150M — generative AI didn't just cut costs, it made the film possible.
Robin Rombach sat with Martin Scorsese multiple times to demonstrate Black Forest Labs' generative tools. What captivated Scorsese wasn't automation — it was the ability to take a visual scene living in his imagination and externalize it for his team to iterate on. Language is lossy. Images are not.
Cerebras is sitting on $25 billion in backlog, and every hyperscaler from OpenAI to AWS faces the same problem: demand is fully booked, and they're racing to keep customers from leaving — not chasing speculative future adoption.
The open source AI landscape has quietly become a geopolitical flashpoint. Outside of OpenAI's OSS model, most available open source options are Chinese. Regulated industries in finance and healthcare that need on-premise, sovereignty-friendly AI have almost nowhere to turn for domestic alternatives.
Analysis
What they talk about
- Technology 57%
- Business 29%
- Government 7%
- Health & Fitness 7%
Connections
Shows they appear on and people they share episodes with. Drag to explore.
