Nvidia, OpenAI Pledge US$20B to AI Chip Startups; Samsung Faces TSMC Pushback
Nvidia and OpenAI each pledge US$20B to back early-stage AI chip startups, aiming to accelerate development of next-generation accelerators. Samsung is leveraging its high-bandwidth memory capacity to secure Nvidia’s upcoming LPU orders, but TSMC is pushing back through capacity expansions and contract revisions.
1. Nvidia and OpenAI Commit $20B to AI Chip Startups
Nvidia plans a US$20B investment fund to back a range of AI chip startups over the next three years, targeting firms developing custom accelerators, novel architectures, and advanced IP. OpenAI is matching this commitment with its own US$20B program designed to secure diversified silicon sources and reduce reliance on incumbent foundries.
2. Samsung Leverages HBM to Secure Nvidia LPU Orders
Samsung is promoting its leading high-bandwidth memory production to win early orders for Nvidia’s upcoming LPU accelerator series, offering integrated supply deals and optimization support. In response, TSMC is expanding capacity at its most advanced nodes and renegotiating terms with Nvidia to protect its share of high-performance compute orders, setting up a direct supply-chain battle.