OpenAI Jalapeño explained in 11min..

OpenAI drops preliminary benchmark for their new Jalapeno chip that shows strong numbers on inference work even compared to NVIDIA Blackwell chips. Semiconductor industries are notorious for being constrained and to only spend 13 months to go from architecture design to production is a pace that I'm certainly impressed across the board given their inclusion of HBM4 in their ASIC as more and more companies like Anthropic, xAI, Meta, and Microsoft all try to develop their own version of customized inference chips. #openai #semiconductor #artificialintelligence Follow me: X: https://x.com/calebfoundry LinkedIn: https://www.linkedin.com/in/calebeom/ TikTok: https://www.tiktok.com/@calebwritescode Chapters 00:00 Intro 00:28 Benchmark 01:10 ASIC 02:16 Performance 03:05 Drawbacks 05:17 InferenceX 06:41 Semiconductors 07:12 Scaling 09:24 Kernels 10:19 Conclusion