OpenAI has begun deploying its Jalapeño ASICs alongside AMD EPYC Turin host CPUs with 1.5TB of memory per host, Hardware VP Richard Ho told Tom's Hardware in an interview published Friday. Ho called the Turin choice pragmatic, saying Nvidia's new Vera CPU was a little bit behind on that maturity level for a fast, de-risked rack design even though OpenAI also works closely with Nvidia.

The comments underline how frontier labs now mix custom accelerators with mature x86 hosts rather than waiting for every new Arm server CPU. SemiAnalysis had described the rack-scale Jalapeño layout after the chip's reveal; Ho's on-record interview confirms Turin is in the live design. Partners' prior Turin experience and socketed serviceability also factored in, he said.

Separately, Nvidia introduced a 64GB DGX Spark configuration built on the GB10 Superchip, with systems from Acer, Asus, Dell, Gigabyte, HP, and MSI slated to start at $4,999 when they launch Oct. 23. Tom's Hardware said the cut from 128GB aims at local AI users who do not need maximum memory amid high RAM prices, while keeping clustering options via ConnectX-7 networking. 128GB Sparks remain available at higher street prices.

Backstory: Meta has also pushed its own Iris AI chip toward production on earlier Reuters reporting, part of a broader push by hyperscalers and labs to diversify beyond a single GPU vendor. Equity traders still treat Nvidia as the bellwether, but CapEx and host-CPU choices increasingly decide who can ship capacity on time.

Where to watch: partner system ship dates for Spark 64GB and any OpenAI comments on Vera or Arm AGI follow-on hosts.

What we know:
OpenAI confirmed Jalapeño racks with AMD Turin hosts. Nvidia set a $4,999 starting point for 64GB DGX Spark systems due Oct. 23.

What we don't:
Production volumes, power and cost-per-token claims versus Nvidia GPUs, and whether Jalapeño later moves onto Vera or other Arm hosts.