OpenAI deploys Jalapeño ASICs with AMD EPYC Turin CPUs, not Nvidia Vera
OpenAI is deploying its new Jalapeño ASIC chips internally alongside AMD EPYC Turin host CPUs, each equipped with 1.5TB of memory. Richard Ho, OpenAI's VP and Head of Hardware, told Tom's Hardware Premium the choice was pragmatic, citing Nvidia's Vera CPU as "a little bit behind… on that maturity level" for a rack-scale deployment. Ho said OpenAI wanted to de-risk the Jalapeño program and move…
Key points
- OpenAI pairs Jalapeño ASICs with AMD EPYC Turin hosts with 1.5TB memory each
- Richard Ho calls Nvidia Vera "a little bit behind… on that maturity level"
- OpenAI is an Arm AGI partner and Vera explorer but chose Turin to de-risk deployment
Nvidia has claimed Vera offers a 1.8x improvement over AMD's EPYC 9755, though that figure comes from an overall benchmark suite where Vera was only 3% ahead. Arm's AGI CPU, also positioned for agentic workloads, claims over twice the performance of modern x86 platforms based on internal estimates. OpenAI is listed as both an Arm deployment partner for AGI and a customer exploring Vera, but Ho indicated the socketed design of Turin and the deeper x86 software foundation factored into the decision. OpenAI may pair Jalapeño with different hosts in the future.
The story so far
2 episodes →- OpenAI deploys Jalapeño ASICs with AMD EPYC Turin CPUs, not Nvidia Verathis story
OpenAI’s Jalapeño ASICs are deployed alongside AMD EPYC ‘Turin’ CPUs as hosts, not Nvidia's Vera
Tom's Hardware · 2 October 2026
Loading the full article…
This text was published by Tom's Hardware and written by Jake Roach. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Hardware & Compute
All →- Sony adds QSSR AI upscaling to PS5 via AMD collaboration · 3 src
- NVIDIA launches DGX Spark 64GB with 100-billion-parameter model support · 4 src
- SpaceX launches Google AI chips into orbit on Falcon 9 rocket · 9 src
- Google shifts AI workloads to TPUs, cutting Nvidia dependency · 1 src
- Apex Compute builds open-source Mesa Vulkan driver for its hardware · 1 src
Comments
via GitHub Discussions