OpenAI and Broadcom Announce 10 Gigawatts of Custom AI Accelerators for Next-Gen AI Clusters

OpenAI & Broadcom Partnership

• 🤝 OpenAI and Broadcom have formed a multi-year partnership to develop and deploy an incredible 10 gigawatts of custom AI accelerators.
• 💡 This collaboration aims to revolutionize the AI landscape by combining OpenAI's cutting-edge knowledge with Broadcom's networking expertise.

Custom AI Accelerators & Hardware Design

• 🧠 OpenAI will design its own chips and systems, directly embedding its insights into the hardware to unlock unprecedented capabilities.
• 🚀 This innovative approach pushes the boundaries of AI achievement by integrating learnings from frontier models directly into the accelerator design.

Deployment & Infrastructure

• 🌐 Racks of these AI accelerators and network systems will be strategically placed across OpenAI's facilities and partner data centers.
• 🔌 Broadcom's expertise in Ethernet and connectivity solutions will be crucial for scaling these next-generation AI clusters.
• 🗓️ Deployments are planned to start in late 2026 and complete by 2029, meeting surging global demand for AI technology.

Vision for AI Advancement

• 🎯 Sam Altman emphasizes this collaboration is crucial for building infrastructure to unleash AI's full potential and benefit humanity.
• ✨ Hock Tan highlights this as a pivotal moment in the pursuit of Artificial General Intelligence (AGI).
• 📈 Charlie Kawwas notes that custom accelerators combined with standards-based Ethernet offer an optimized, cost-effective, and power-efficient AI infrastructure.

Impact & Future Outlook

• ✅ OpenAI's growth, with over 800 million weekly active users, positions it to lead in ensuring AGI benefits everyone.
• 🔑 This partnership sets new industry standards for open, scalable, and power-efficient AI clusters, shaping the future of AI.