Meta has signed a multi-year agreement with Advanced Micro Devices (AMD) to deploy up to 6 gigawatts of AI GPUs, in one of the largest infrastructure bets in the field of artificial intelligence. The company will start rolling out the chips from the second half of 2026 across its global data-center network.
The deal marks a shift in how Big Tech measures computing capacity. Instead of the number of processors, companies now talk in terms of power consumption. A 6-gigawatt deployment can support hundreds of thousands of high-end accelerators and requires multiple hyperscale facilities.
Meta is expanding compute to train large AI models and run AI agents across Facebook, Instagram, and WhatsApp. The company has already guided for massive capital expenditure on AI infrastructure this year, signaling an aggressive scale-up.
Meta now has its second major supplier for advanced AI chips through this partnership. Meta distributes its orders to multiple vendors while supporting its custom silicon development because this strategy helps it prevent supply issues and manage its expenses.
The contract provides AMD with its main hyperscaler customer while it boosts the company’s position in high-performance AI markets. The announcement increased AMD’s stock price during initial trading because investors trusted the company’s data center strategic plan.
Also Read: NVIDIA Eyes Mass-Market Laptops With AI Chips, Taking Fight to Intel, AMD
The agreement comes at a time when global technology firms are committing record sums to AI infrastructure. Long-term supply pacts now double as capacity-locking mechanisms in a market where demand for advanced chips continues to outstrip supply.
Analysts say such power-linked deals show that leadership in AI will depend as much on access to compute as on model innovation. The Meta-AMD pact, they add, secures future capacity while accelerating the company’s push to build next-generation AI at scale.