Making Markets · Colossus | Investing & Business Podcasts

Paul Bloch: Powering the AI Revolution - [Making Markets, EP.68]

·39 min·2 clips

Bloch says NVIDIA, XAI, and DDN are now tied to a race for speed, scale, and AI deployment.

The premise is blunt. The guest thinks even conservative people will be pushed toward AI eventually, so the choice is less whether to adopt it and more whether to arrive early or late. His internet comparison is not sentimental. It is about how fast people learn once a new medium starts making sense. Email and file transfer changed expectations over time. AI, in his telling, moves that pattern faster and with a much higher ceiling. Government is one of the clearest examples. The guest expects every country to want some form of AI factory tied to sovereign AI policy. Bureaucracy is the easy case to picture: passport documents, personal documents, travel, car information, taxes, and other citizen data could move through a simpler interface. Then he pushes the idea past public services. Every vertical inside a country has local talent and local data, and the question is how that turns into useful output. The conversation gets more grounded when it moves to the physical layer. Using high-frequency trading as a comparison, it asks whether location and speed still matter when models work with huge stores of data. The answer is not clean, which is probably the honest part. More compute matters, but locality still matters because GPU nodes, networks, and storage have to work together. Moving data is not just a line item. By the end, AI sounds less like software magic and more like a capital-heavy infrastructure race. Storage, speed, geography, and capital allocation all show up. The labor point is sharper than the usual replacement talk: money has to come from somewhere, output has to rise somewhere, and people still need to provide service or add value as automation raises the baseline.

Listen to the show on

Apple Podcasts Spotify

Play clips from this episode More from Making Markets