Now Playing
Anthropic's 2nm ASIC, Intel's Price Hike & Meta's GPU Rental Play
Audio is available on Spreaker — see link below.
Anthropic confirms Samsung 2nm inference chip talks as the ASIC race accelerates past OpenAI's Jalapeño launch — custom silicon is now a competitive necessity. Plus: Nvidia's export control credibility erodes, Intel hikes Arrow Lake prices on a dead-end socket, and Meta opens its GPU clusters to paying cloud customers.
Audio is available on Spreaker — see link below.
Anthropic is in talks with Samsung to build a custom two-nanometer inference accelerator, and the timing tells you almost everything about where the AI hardware race stands right now. The confirmation came through a combination of engineering headcount and direct reporting.
What makes this moment specific is OpenAI. The unveiling of OpenAI's Jalapeño chip in late June compressed the competitive timeline sharply.
Elsewhere, Nvidia's policy credibility is taking damage. The company submitted arguments to the Trump administration suggesting Huawei could satisfy global AI chip demand if export controls remained in place.
On the consumer CPU side, Intel's Arrow Lake Refresh launched and then immediately got more expensive. Prices on the Core Ultra two-seventy-K Plus and two-fifty-K Plus rose fifteen to seventeen percent, a fifty dollar increase, shortly after launch.
Meta's move is worth watching on a different axis. The company is shifting its H100 and B200 clusters from internal-only use toward a cloud rental model, effectively competing with AWS, Google Cloud, and Azure on AI compute pricing.
Chapter summary auto-generated from the verified script. Listen to the full episode for the complete content.