Nvidia and AMD Invest $4 Billion in Meta's $97.5 Billion AI Stock Offering as Chip Partnerships Deepen

Meta Platforms upsized its equity fundraising program to $97.5 billion, drawing anchor investments from its two biggest chip suppliers: Nvidia will put $3 billion into the offering and AMD will contribute $1 billion, with both companies simultaneously expanding their AI hardware partnerships with the social media giant. The combined $4 billion in chip-company commitments came alongside detailed technology agreements centered on Meta's Muse AI agent platform, raising questions on Wall Street about the circularity of the financing structure.

Nvidia's deal pairs the equity investment with a broad expansion of Meta's use of Nvidia hardware. Meta plans to increase deployments of Nvidia Vera Rubin systems and Vera CPUs, along with Spectrum-X Ethernet networking and BlueField-4 data processing units, to power next-generation AI agents. The two companies are also jointly optimizing Muse model training and inference workflows to reduce total cost of ownership, and are exploring collaboration on Nvidia's OpenShell reference architecture for safer AI agent deployment and the Nvidia Sentry platform for monitoring agent behavior. Nvidia also separately disclosed it is extending the useful lifespan of its GPUs to eight years, a move analysts say could reshape AI asset valuation and financing strategies.

AMD's $1 billion investment came alongside confirmation that its first Helios rack-scale AI systems are installed at Meta and undergoing testing ahead of broader deployment. Meta already runs millions of AMD EPYC processors, including those powering Muse, and uses Instinct GPUs for AI training, ranking, and recommendation workloads. Looking ahead, Meta plans to deploy a custom MI450-based Instinct GPU optimized for its workloads and is a lead customer for AMD's 6th Generation EPYC processors. The companies are also deepening software collaboration across ROCm and PyTorch, building on a previously announced 6-gigawatt AMD GPU agreement.

Meta's overall equity program initially targeted $35 billion in underwritten public offerings before being enlarged to $37.5 billion in common stock and depositary shares. A further at-the-market program of up to $60 billion is planned for 2027, bringing the combined potential to $97.5 billion, all directed at funding AI infrastructure. AMD shares gained more than 4% on the news, Nvidia edged up about 0.7% in premarket trading, while Meta fell roughly 1.9% amid investor concern over the scale of capital expenditure the program implies.

Separately, Oracle held its AI World event in Las Vegas and announced a sweeping expansion of its cloud chip lineup. Nvidia Vera Rubin NVL72 systems and Vera CPUs are now generally available on Oracle Cloud Infrastructure, with the partnership targeting multi-gigawatt scale. Oracle also said it will offer Google TPUs on its platform starting in 2027, making it the first hyperscaler outside Google to do so, and added Cerebras fast-inference chips and SambaNova's SambaRack SN50 systems to its cloud catalog for 2027. Intel and Oracle announced a multiyear collaboration supporting enterprise and agentic AI workloads. On the software side, Oracle expanded its partnership with OpenAI to integrate ChatGPT and Codex across Oracle applications, cybersecurity, and cloud services, and said it will offer Snowflake's AI Data Cloud as a native first-party service on Oracle Cloud. Oracle shares rose about 4 to 5 percent on the announcements.