Meta set the stage on Monday with plans to raise $95 billion for AI infrastructure: $35 billion through underwritten public equity offerings and another $60 billion via an at-the-market program. The scale underscored how sharply the race for data-center capacity is reshaping Big Tech’s balance sheets.¹

By Tuesday, Nvidia had committed $3 billion to the offering and AMD another $1 billion, turning Meta’s fundraising into a supply-chain alliance as much as a capital raise. The companies had previously been named, alongside Broadcom, as prospective buyers of up to $5 billion in Meta shares.¹

For Meta, the pitch is bigger than chips. The company said its partnership with Nvidia would expand work on Muse and its family of models, aimed at improving training and inference performance. Mark Zuckerberg framed the effort as a push to bring “the best personal agents and frontier models” to billions of people.²

Nvidia, meanwhile, casts the partnership as a change in the basic relationship between people and software. “AI agents represent a new computing paradigm — software that understands our goals, reasons, uses tools and gets work done on our behalf,” CEO Jensen Huang said, arguing that users will increasingly work with intelligent agents rather than simply operate applications.²

AMD’s case is more explicitly industrial. It said its Helios racks are already running at Meta and its EPYC processors power Muse. CEO Lisa Su said the companies were building infrastructure “at gigawatt scale” while shaping future silicon, software and systems for Meta’s superintelligence ambitions.¹

The arrangement gives Meta immediate firepower for an enormous expansion. It also gives its chip suppliers something equally valuable: a clearer claim on the future hardware demand generated by that expansion.