Silicon Futures for August 2026 – Hot Chips 2026, Octane, OpenAI Jalapeño

Silicon Futures is a neXt Curve reThink Podcast series on AI and semiconductor tech and the industry topics that matter.

This month, Leonard Lee (neXt Curve), Karl Freund (Cambrian AI Research), and Jim McGregor (Tirias Research) unpack the hottest takes from Hot Chips 2026 and the incessant drama that is AI & AI infrastructure!

Are we on the verge of a major shift in AI HPC computing as the AI focus shifts from training toward inference and agentic AI?

What is the impact of the broadening memory crisis and innovations in sparse MOE models on the future of AI computing architectures?

What does the introduction of OpenAI’s Jalapeño mean for the future of AI HPC architectures and chips?

We break down the AI and chip trends that matter! Tune in!

In this episode, Leonard, Karl and Jim talk about….

➡️ IBM’s dual-ISA core with native Arm + zArchitecture (4:54)
➡️ How will HRL Labs impact IBM’s quantum leadership? (8:11)
➡️ What were the key themes of Hot Chips 2026? (10:13)
➡️ Should Optane be resurrected for AI HPC? (12:50)
➡️ 3D-memory stacking & PIM (Processing in Memory) (13:24)
➡️ HBF: What is it good for? Fad or frontier? (13:45)
➡️ The economics & margins of the memory industry! (15:17)
➡️ Our takes on NVIDIA’s Q2 2027 earnings results! (20:20)
➡️ Bill Gates AI essay and physical AI in context… (29:49)
➡️ For AI data centers, can coding be enough? (32:02)
➡️ NVIDIA partners with D-Matrix on inference & agentic! (33:31)
➡️ Why do we compare AI scaling with Moore’s Law? (37:38)
➡️ Is tokenomics the only thing that matters? (39:44)
➡️ Is the AI ROI debate over or early days? (40:52)
➡️ Beyond disaggregation & distributed AI inference! (43:20)
➡️ OpenAI tapes out Jalapeño from RTL in 9 months! (45:50)
➡️ Did DeepSeek change everything or nothing? (54:30)