Meta Platforms is significantly expanding its partnership with Nvidia, committing to deploy millions of the chipmaker’s advanced GPUs and, notably, standalone CPUs within its artificial intelligence data centers. The deal, likely worth tens of billions of dollars, underscores Meta’s aggressive investment in AI infrastructure as it aims to deliver what CEO Mark Zuckerberg calls “personal superintelligence” to its billions of users. The expansion was announced on .
While Meta has utilized Nvidia’s graphics processing units for over a decade, this latest agreement represents a broadening of the technological collaboration. Crucially, Meta will be the first company to deploy Nvidia’s Grace central processing units as standalone chips in its data centers, moving beyond the traditional model of integrating CPUs alongside GPUs on a server. This shift signals a strategic bet on Nvidia’s CPU capabilities and a potential disruption to the dominance of Intel and AMD in the server CPU market.
The financial terms of the deal remain undisclosed, but analysts estimate it to be in the “tens of billions of dollars,” representing a substantial portion of Meta’s planned capital expenditure of up to $135 billion for AI in . The commitment comes as Meta intensifies its focus on AI-driven features across its platforms, including WhatsApp, and seeks to enhance its personalization and recommendation systems.
Nvidia’s offerings in the deal extend beyond CPUs and GPUs to include networking technology, essential for managing the massive data flows required for AI training and inference. Meta will leverage Nvidia’s Blackwell and Rubin GPUs, as well as its networking solutions, within its data centers. The company also plans to roll out Vera CPU-only systems in , offering a more traditional server architecture alongside its GPU-accelerated infrastructure.
The expanded partnership has already impacted market sentiment. Shares of both Meta and Nvidia climbed during extended trading on , while Advanced Micro Devices (AMD) stock experienced a decline of approximately 4% on the news. This reflects investor concern that Meta’s increased reliance on Nvidia could diminish opportunities for AMD and Intel in the data center space.
“No one deploys AI at Meta’s scale — integrating frontier research with industrial-scale infrastructure to power the world’s largest personalization and recommendation systems for billions of users,” stated Nvidia CEO Jensen Huang. This highlights the scale of Meta’s AI ambitions and the critical role Nvidia’s technology plays in enabling them.
The deal also extends to Nvidia’s cloud partner program, allowing Meta to access chips hosted by companies like CoreWeave and Crusoe, providing flexibility and scalability in its AI infrastructure deployment. This approach allows Meta to leverage external resources to supplement its own data center capacity.
While Meta continues to develop its own in-house silicon, the Nvidia partnership demonstrates a strategic decision to diversify its chip sourcing and leverage the leading-edge technology offered by Nvidia. The company had previously considered utilizing Google’s Tensor Processing Units (TPUs) as an alternative, but has now solidified its commitment to Nvidia for the foreseeable future.
The move towards standalone CPUs is particularly noteworthy. Traditional server designs typically integrate CPUs and GPUs on the same chip or board. Deploying CPUs independently allows Meta to optimize its infrastructure for specific AI workloads and potentially achieve greater efficiency. This could also pave the way for more specialized server designs tailored to the demands of AI applications.
The implications of this deal extend beyond Meta and Nvidia. It signals a broader trend of increasing investment in AI infrastructure and the growing importance of specialized hardware in powering the next generation of AI applications. The competition among chipmakers to secure contracts with major cloud providers and AI developers is expected to intensify as the demand for AI processing power continues to surge.
The partnership is not without potential risks. Reliance on a single vendor, even a leading one like Nvidia, can create supply chain vulnerabilities and limit negotiating power. However, Meta appears to be mitigating this risk by continuing to invest in its own chip development efforts and exploring alternative options. The long-term success of this strategy will depend on Meta’s ability to effectively manage its chip sourcing and maintain a competitive edge in the rapidly evolving AI landscape.
