Meta strikes AI chip deal with AMD days after committing to deploy millions of Nvidia GPUs

Meta announced on Tuesday a multiyear deal with Advanced Micro Devices to deploy up to 6 gigawatts of AMD's graphics processing units for AI data centers, just one week after committing to use millions of Nvidia processors for its AI expansion, as the social media giant invests up to $135 billion in capital expenditures this year to compete in the global AI race. The agreement includes AI-optimized central processing units and will see early shipments of MI450 GPUs in AMD's Helios rack-scale servers begin later this year. AMD CEO Lisa Su emphasized that this deal places her company "at the center of the global AI buildout" by delivering "high-performance, energy-efficient infrastructure optimized for Meta's workloads." This strategic partnership represents a significant development for AMD, which currently lags far behind Nvidia in the competitive AI chip market. While Nvidia now stands as the world's largest publicly traded company with a $4.66 trillion valuation and approximately 90% market control, AMD is valued at $320 billion. The deal with Meta is particularly notable as it involves customized GPUs, a differentiator from Meta's agreement with Nvidia according to chip analyst Ben Bajarin of Creative Strategies, who estimates the contract could be worth tens of billions of dollars over at least four years. Meta's decision to diversify its AI chip suppliers comes amid intense competition in the AI space, with the company racing against megacap peers and AI firms like OpenAI and Anthropic. Meta's aggressive chip procurement strategy reflects its broader commitment to expanding AI capabilities across its platforms. The company plans to develop 30 data centers, including 26 in the United States, to support its AI initiatives. This latest agreement follows Meta's established pattern of working with multiple chip suppliers, including both AMD and Nvidia, while also developing in-house processors. Notably, Meta had reportedly been in discussions with Google about potentially deploying Google's tensor processing units in its data centers by 2027. The timing of AMD's announcement, just before Nvidia's quarterly earnings report where analysts project 68% revenue growth to $66 billion, underscores the intensifying competition in the AI chip market as companies seek to secure critical infrastructure for next-generation artificial intelligence systems.