OpenAI's Chip Strategy: Winning Billions from Nvidia Even If the Chip Fails
AccBalanced · x · 2026-08-26
SemiAnalysis analyzes OpenAI's internal 'Jalapeño' chip, arguing OpenAI wins regardless of deployment scale. By launching an ASIC program and presenting at Hot Chips, they pressure Nvidia to secure billions in funding and backstops. Compared to Cerebras, an 8-chip Jalapeño node holds 1.7TB memory (enough for a 1.6T param model), requiring significantly less CapEx and power (11kW) than Cerebras' wafer-scale solution ($20M+ CAPEX, 1MW).
Related event: OpenAI's Jalapeño Chip Play: Winning Either Way(2 posts)→
More from Infra
- OpenAI defines three phases of disaggregated compute — beffjezos · 2026-08-26
- Jalapeno introduces latencies not seen before in HBM-based solutions — BenBajarin · 2026-08-26
- OpenAI reveals 'Jalapeno' inference ASIC at Hot Chips — beffjezos · 2026-08-26
- OpenAI's custom chip 'Jalapeño' reportedly beats Nvidia Blackwell in efficiency — yacineMTB · 2026-08-26
- Google's Hot Chips Talk Praised for Technical Depth — firstadopter · 2026-08-26
- Google exec calls out HBM memory as critical bottleneck — firstadopter · 2026-08-26