OpenAI's Jalapeño Chip Outperforms Nvidia Superchips
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- OpenAI published a blog post Tuesday claiming its Jalapeño chip completes AI tasks more efficiently with lower latency and higher throughput, addressing what hardware VP Richard Ho called the typical trade-off between the two.
- Jalapeño delivered 1.5 to 1.9 times more AI work per watt and 1.7 to 3.6 times lower end-to-end latency than Nvidia's GB200 and GB300 superchips across GPT-OSS 120B, DeepSeek R1, and Kimi K2.5 1T on the InferenceX benchmark.
- Jalapeño is an Application-Specific Integrated Circuit (ASIC) built in partnership with Broadcom, first introduced in June and designed specifically for AI inference workloads.
- OpenAI plans to deploy Jalapeño in small volumes by the end of this year and ramp up production into 2027, while developing second and third generations of the chip.
- OpenAI does not plan to replace its entire chip lineup with Jalapeño — Ho said the company's compute strategy includes 'very good partners' like Nvidia, signaling continued coexistence rather than displacement.
Why it matters: OpenAI's published 1.5–1.9x efficiency gains and 1.7–3.6x latency cuts over Nvidia's flagship superchips give the company a custom-silicon alternative for inference at scale, with limited volumes shipping by end of 2025 — but Ho's explicit reassurance that Nvidia stays a 'very good partner' shows this is leverage, not a breakup.
Ask SkimNews
