New pod episode with Austin Lyons where we discuss all things Nvidia at CES 2026!

- Where was the Infiniband switch? "6 chips" instead of 7

- Uneducated investor panic and Jensen's Midas touch

- DGX Spark (GB10) demo was awesome. MacBook as the thin client, DGX Spark as the local AI server. Maybe DGX Spark is the real AI PC?

- DGX Station (GB300) moves frontier AI compute from the cloud and makes it accessible on-prem in the enterprise. Still 1600W though (and maybe liquid-cooled?) so maybe not at your desk?

- The historical parallel is $1M+ mainframes vs $100K minicomputers, expanding compute TAM rather than displacing cloud usage.

- Context memory storage elevates SSDs, enabling massive KV cache offload beyond HBM and DRAM. Similar tech already exists (Supermicro JBOF)

- What's the DPU? Compute in the network

- The new Vera CPU highlights rising CPU importance for agentic workloads through higher core counts, SMT, and large LPDDR capacity.