OpenAI launches Jalapeño AI chip with Broadcom, challenging Nvidia's dominance

OpenAI recently introduced its first custom artificial intelligence (AI) processor, dubbed Jalapeño, marking a major move by the ChatGPT-maker to develop in-house silicon and mounting new pressure on Nvidia’s near-monopoly over advanced AI computing hardware. The custom-built inference semiconductor is claimed to deliver industry-leading speed and operational efficiency, positioning OpenAI alongside tech titans like Google, Amazon Web Services (AWS), Microsoft and Meta in engineering proprietary chips to power massive artificial intelligence systems. Developed in collaboration with semiconductor giant Broadcom, the Jalapeño processor is specifically engineered to handle inference, the phase where trained models execute tasks and process user queries. OpenAI plans to deploy Jalapeño across its computing infrastructure before the end of the year, confirming that engineering teams are already working on second- and third-generation versions. How Jalapeño stacks up against Nvidia hardware While Nvidia’s market valuation has soared throughout the global data centre expansion due to increasing demand for its GPUs across model training and day-to-day inference. However, market analysts warn that the rapid emergence of hyperscaler-designed silicon presents a growing threat to Nvidia’s long-term dominance, particularly within the fast-growing inference market. Evaluating the chip’s performance inside OpenAI’s development facilities, independent research firm SemiAnalysis noted that Jalapeño surpassed Nvidia's Blackwell architecture in performance per watt across nearly every testing benchmark. For example, analysts noted that the direct comparison to Blackwell is somewhat asymmetric because Jalapeño incorporates next-generation HBM4 memory, making Nvidia’s forthcoming Rubin platform a more accurate peer comparison. TrendForce analyst Fion Chiu observed that while Jalapeño will decrease OpenAI’s reliance on Nvidia for daily inference tasks, Nvidia’s GPUs will remain indispensable for large-scale model training and frontier AI workloads due to their flexible programmability and established CUDA software ecosystem. Tech industry’s shift toward custom chips OpenAI’s silicon debut reflects an accelerating transition across Big Tech to deploy custom application-specific integrated circuits (ASICs). First to the party was Google with its TPU expansion, Google continues to roll out next-generation Tensor Processing Units (TPUs) across its cloud network for both model training and live inference. Meta joined the fray after it formalised agreements to deploy one gigawatt of Broadcom-engineered custom AI processors as part of a multi-GW infrastructure plan. Anthropic committed over $100 billion over the next decade toward AWS infrastructure, including Amazon’s proprietary Trainium processors. Meanwhile, emerging chipmakers such as Cerebras, SambaNova, D-Matrix, Etched and Fractile are similarly advancing specialised AI accelerators.