In this episode, Kevin Bowers returns to explain Jump's expansion from trading to core infrastructure, centered on what he calls the "great inversion": the real bottleneck in tech isn't compute, but data and I/O. He introduces Shelby, a new storage network, as a direct challenge to the "Hotel California for Data" model used by cloud providers. This same focus on efficient data flow—not just processing power—was the key to scaling Solana with Fire Dancer. Finally, Kevin explains how FPGAs from high-frequency trading are the critical hardware solution, allowing blockchains to bypass software's inefficient "Tower of Babel" and "get close to the wire" for true high performance.

00:00 - Expanding Beyond Trading and the Vision for Shelby

02:31 - Challenges in Storage and Data Management

04:37 - Building High-Performance Systems

08:04 - The Evolution of Jump's Technology

11:55 - The Economics of Cloud Storage

29:07 - Fire Dancer and Frankendancer

42:03 - The Cost of Optimization

42:48 - Machine Learning and Custom Networks

43:47 - Project Prioritization and Entropy

46:48 - Challenges in High-Performance Computing

56:38 - The Role of FPGAs in Trading and Blockchain

01:13:45 - Future of Hardware Acceleration in Blockchain

01:18:54 -Conclusion and Final Thoughts