Major companies are designing specialized processors for AI inference, such as OpenAI's Jalapeño with Broadcom and Qualcomm's Dragonfly, challenging Nvidia's dominance. This shift towards custom silicon promises better efficiency and cost-effectiveness for AI workloads.
Nvidia is betting on co-packaged optics as a lever for the next generation of AI infrastructure. At the heart of the roadmap is TSMC’s COUPE platform, integrating photonics and silicon for more efficient interconnects. For organizations managing on-p...
OpenAI’s new inference chip, designed by Broadcom and manufactured by TSMC, marks the model maker’s entry into the custom silicon race. The move signals a shift in the AI hardware supply chain and hints at a future where on-premise deployments could ...
Qualcomm enters the data center arena with its Dragonfly lineup and a Meta partnership, aiming to challenge Nvidia's dominance. A signal for those seeking AI hardware alternatives.
OpenAI and Broadcom announce Jalapeño, a custom chip designed for Large Language Model inference in data centers. The first step in a long-term roadmap, it aims to reshape efficiency and costs for large-scale inference. AI-RADAR examines the implicat...
At its 2026 Investor Day, Qualcomm unveiled the Dragonfly C1000 CPU, a proprietary high-bandwidth memory alternative to HBM, and new AI accelerators. The push into data center racks signals a shift that could reshape the landscape for those seeking h...
Broadcom and OpenAI have unveiled Jalapeño, OpenAI's first custom processor. It's a massive reticle-sized ASIC optimized exclusively for inference, built in an ultra-fast nine-month cycle. The move underscores the race toward proprietary silicon for ...
OpenAI and Broadcom's collaboration yields a custom processor optimized for large language model inference. The Jalapeño announcement marks an acceleration toward specialized hardware, with potential impact on efficiency and data sovereignty for ente...
