Jalapeño shows 50% lower inference cost per token vs Nvidia in early testing — Broadcom CEO Hock Tan personally delivered engineering samples to Sam Altman + Greg Brockman at OpenAI HQ
Early lab testing shows Jalapeño delivers approximately 50% lower inference cost per token than current-generation Nvidia GPUs — with performance matching Nvidia Blackwell and Google TPUs. Broadcom President + CEO Hock Tan personally delivered engineering samples to OpenAI CEO Sam Altman + President Greg Brockman at OpenAI San Francisco headquarters. The cost-economics + delivery-symbolism specifics validate the announcement substantively.
The substantive piece is the 50%-cost-reduction operational economics specifics. Yesterday's Jalapeño announcement characterized 'better perf-per-watt' qualitatively. Today's 50% lower inference cost per token specification + performance-matching-Blackwell+TPUs framing provides quantitative operational economics that procurement-evaluation teams can model against.
The competitive read for Nvidia's structural pricing power is that 50% inference-cost reduction at performance parity represents substantive pricing-pressure threshold. Combined with AMD's 6GW multi-year OpenAI supply, OpenAI's non-Nvidia silicon strategy substantially affects Nvidia's pricing-power baseline through H2 2026 to 2030 inference-deployment landscape.
Build Fast With AI — AI News Today June 26 2026 → · AI Tools Recap — AI News June 26 2026 →