Introduction
Advanced Micro Devices (AMD) has seen sharp volatility in recent weeks, with the stock down about 18% over the past month while the SPDR S&P 500 ETF (SPY) edged higher over the same stretch. The decline accelerated after AMD’s latest quarterly update, yet the narrative shifted again after a major expansion of its AI infrastructure partnership with Meta Platforms. The key question for investors is whether a large multi-year AI deployment can offset near-term concerns about guidance, competition, and dilution.
Why AMD Sold Off After Earnings
The stock drop intensified following AMD’s Q4 earnings report on February 3. AMD closed at $242.11 and then closed at $200.19 the next day, a decline of about 17% in a single session. The primary driver cited for the selloff was disappointing guidance, which weighed on sentiment even though AMD’s longer-term price performance has remained strong.
Zooming out, the contrast is notable. Over the past year, AMD is up roughly 104% versus about 16% for SPY. That gap reflects sustained enthusiasm for AMD’s positioning in AI and data center compute, even as short-term expectations can swing sharply around earnings and forward outlook.
Meta Expands the Partnership to 6GW of AI GPUs
On February 24, AMD and Meta announced an expanded partnership aimed at powering Meta’s AI infrastructure with up to 6 gigawatts of AMD Instinct GPUs, paired with AMD EPYC CPUs. AMD highlighted the agreement as a multi-year, multi-generation collaboration spanning GPUs, CPUs, and rack-scale AI systems designed around Meta’s workloads. Meta also indicated it expects to be a lead customer for the next-generation EPYC Venice CPU.
A core feature of the structure is a performance-based warrant issued by AMD to Meta for up to 160 million shares, designed to vest as shipment milestones are met. This approach mirrors a prior incentive structure AMD used in another high-profile AI deal, reflecting the challenge of winning deployments in a market still heavily oriented around Nvidia’s software ecosystem.
Bank of America Sees 2030 EPS Upside, With Caveats
Bank of America analyst Vivek Arya and his team argued the Meta agreement could be meaningfully accretive to AMD’s longer-term earnings power. They estimated that each gigawatt of deployment could represent roughly $15 billion to $20 billion in net revenue opportunity for AMD, translating to about $6 billion in net income opportunity. On that basis, the team projected a potential 13% to 23% boost to AMD’s EPS by 2030, even after accounting for up to roughly 10% potential dilution tied to the warrant structure.
At the same time, the equity component raises questions about the quality and durability of the economics. The structure can be viewed as a form of circular incentive that helps catalyze adoption, but it also shifts part of the value to the customer and introduces dilution sensitivity. Arya’s view was that the stock would need to rise substantially for full dilution to be realized, which could make the tradeoff more tolerable for shareholders if the business impact scales as expected.
Competitive Context and Execution Risks
Strategically, AMD is trying to expand beyond the perception that Nvidia is the default choice for AI accelerators, a challenge often described as a software and developer ecosystem moat. AMD’s rack-scale approach, including its Helios system developed with Meta through the Open Compute Project, is meant to strengthen its full-stack credibility by pairing GPUs, CPUs, and system-level architecture that large buyers can deploy at scale.
Key risks remain. Analysts have highlighted execution on first rack-scale products, dependence on outsourced manufacturing, the lumpy nature of enterprise and consumer spending cycles, and uncertainty around the timing and magnitude of large AI infrastructure commitments. Potential upside includes stronger-than-expected share gains in PC and server processors, along with increasing emphasis on CPUs for inference workloads, which could expand EPYC’s role inside large AI deployments.
Conclusion
AMD’s recent drawdown reflects investor sensitivity to guidance and near-term expectations, but the expanded Meta partnership introduces a substantial long-horizon opportunity tied to AI infrastructure buildouts. The main implications revolve around execution and economics: if AMD can deliver at scale and translate deployments into durable margin and earnings gains, the deal could materially lift long-term EPS power. If not, dilution and competitive pressures could limit shareholder benefits despite headline growth.
