Written by Frode Skar, Finance Journalist.
Meta and AMD announce 6 gigawatt GPU deal as AI arms race intensifies
Meta and AMD secure multiyear 6 gigawatt GPU agreement as capital spending accelerates
Meta and AMD have entered into a multiyear agreement under which Meta will purchase up to 6 gigawatts worth of AI chips, marking one of the largest GPU supply commitments in the current artificial intelligence expansion cycle.
As part of the Meta and AMD agreement, AMD will issue 160 million shares of common stock to Meta. The shares will vest in tranches tied to performance milestones, with the first trigger activated once AMD ships its first 1 gigawatt of chips. The structure directly links operational execution to long term value creation.
AMD shares surged as much as 10 percent in premarket trading following the announcement before moderating during the regular session.
Meta and AMD deepen AI infrastructure build out
The Meta and AMD partnership comes as Meta plans to spend more than 135 billion dollars in capital expenditures in 2026 alone, largely focused on artificial intelligence infrastructure. That includes data center construction, chip procurement and large scale model training.
Under the agreement, AMD’s MI450 GPU line will be deployed in Meta’s Helios rack scale data center systems, paired with EPYC CPUs. Initial shipments are scheduled for the second half of the year.
Meta will also purchase significant volumes of AMD CPUs, including the Venice chip and next generation Verano processors. CPUs are becoming increasingly important in AI data centers as companies shift focus toward inference workloads and agent based AI systems.
The Meta and AMD deal underscores the evolving balance between GPUs, which dominate training tasks, and CPUs, which are critical for running models at scale.
Strategic diversification beyond Nvidia
The Meta and AMD agreement follows a separate multiyear commitment Meta recently made with Nvidia for millions of Blackwell and Rubin GPUs. Nvidia also confirmed that Meta will host large scale deployments of its Grace CPU servers.
By entering into major supply deals with both Nvidia and AMD, Meta is clearly diversifying its chip sourcing strategy. This reduces supply chain concentration risk and strengthens Meta’s negotiating position in a market where demand for AI hardware continues to exceed supply.
The Meta and AMD deal can therefore be viewed as both a capacity expansion move and a strategic hedge.
Capital intensity and investor scrutiny
The scale of the Meta and AMD agreement highlights the extraordinary capital intensity of the AI build out. Industry wide, Meta, Amazon, Google and Microsoft are collectively expected to spend roughly 650 billion dollars on AI infrastructure.
Such spending levels have prompted questions from investors about long term return on capital. Since announcing aggressive AI investment plans, several large technology stocks have experienced volatility.
Meta shares have held up relatively better than some peers, while Microsoft and Amazon have seen sharper pullbacks after outlining their capital expenditure strategies.
Chipmakers have also faced scrutiny. Concerns about a potential AI investment bubble and the development of custom in house chips by major cloud providers have tempered enthusiasm across the semiconductor sector.
Custom silicon versus general purpose GPUs
Reports have indicated that Meta has explored alternative chip options for specific AI workloads. Large technology firms are increasingly developing proprietary accelerators tailored to internal needs.
However, analysts argue that custom chips remain specialized and are unlikely in the near term to fully replace the more flexible and powerful GPUs produced by companies such as AMD and Nvidia.
The Meta and AMD agreement suggests that, at least for now, hyperscale AI infrastructure will continue to rely heavily on high performance general purpose GPU architectures.
Financial implications for AMD
For AMD, the Meta and AMD partnership represents a significant revenue pipeline over multiple years. The performance based equity structure aligns incentives while potentially supporting earnings per share growth if milestones are achieved.
The agreement also strengthens AMD’s competitive positioning in the AI data center segment, where Nvidia has historically held a dominant share.
A guaranteed multiyear commitment from one of the largest AI spenders in the world materially enhances visibility into AMD’s future demand profile.
Conclusion
The Meta and AMD 6 gigawatt GPU agreement marks a major escalation in the global AI infrastructure race.
For Meta, the deal secures critical compute capacity in a market defined by scarcity and rapid scaling requirements. For AMD, it delivers long term revenue visibility and reinforces its status as a core AI hardware supplier.
The Meta and AMD partnership reflects a broader reality: the AI expansion is not slowing. Instead, it is becoming more capital intensive, more competitive and increasingly central to the strategic direction of the world’s largest technology companies.
