OpenAI’s first Jalapeño benchmarks show up to 1.9x higher performance per watt and up to 3.6x lower latency across three AI models.
OpenAI and Broadcom have launched Jalapeño, their first custom AI inference processor built to improve efficiency and reduce reliance on Nvidia chips.
Anthropic has partnered with Broadcom and Google to develop AI chips, strengthening its infrastructure for advanced AI systems.
OpenAI teams up with Broadcom to develop 10 gigawatts of in-house AI accelerators, marking a major move toward self-reliant chip infrastructure for next-generation AI models.
Next-gen AI evolves into “Physical AI”—autonomous factories, wearable exoskeletons, and service robots are embedding intelligence into the real world.
