
AMD reported a 107% surge in data center revenue in Q2 and projects the segment will more than double in 2027, positioning itself as a major player in inference and agentic AI workloads.
Although the stock dipped following the earnings announcement, the company's shift from AI training to inference—which requires a much higher ratio of CPUs to GPUs—and upcoming Helios system deployments with major cloud providers suggest significant growth ahead.
What happened
AMD reported a 107% year-over-year surge in data center revenue to $6.7 billion in Q2, with adjusted earnings per share jumping 246% to $1.66. The company projects data center revenue will more than double in 2027 and guided Q3 revenue to grow 41% year over year to $13 billion (plus or minus $300 million), yet shares fell despite the results.
Why it matters
AMD is positioned to capitalize on a shift from AI training to inference and agentic AI, which require far more CPUs relative to GPUs—changing the GPU-to-CPU ratio from 8-to-1 for training to 1-to-1 for agentic AI. The company projects this represents a $220 billion market opportunity in the next few years, and it has major inference deals with OpenAI, Meta Platforms, Anthropic, and Microsoft.
What to watch
AMD will begin delivering Helios, its first full-rack AI system combining GPUs, CPUs, and networking, this quarter, with shipments expected to ramp up in Q4. The stock is trading at a forward P/E ratio of 35 times 2027 analyst estimates but a forward PEG ratio of only 0.3.
AMD reported strong Q2 earnings that underscore its growing role in AI infrastructure, though the market's immediate reaction was negative. Total revenue climbed 50% year over year to $11.54 billion, accelerating from 38% growth in Q1. Adjusted earnings per share surged 246% to $1.66, beating consensus estimates of $1.62 on $11.28 billion in revenue. The standout performance came from data center, which skyrocketed 107% year over year to $6.7 billion. Client revenue jumped 23% to $3.1 billion, while gaming revenue fell 31% to $779 million due to lower semi-custom sales. The embedded segment grew 19% to $977 million. Adjusted gross margin expanded to 56%, up 1,300 basis points from a year ago, benefiting from the shift toward higher-margin data center products.
The strategic opportunity underpinning AMD's growth centers on inference and agentic AI, which require a fundamentally different hardware profile than the training phase that dominated 2023–2024. During training, the GPU-to-CPU ratio stands at 8-to-1; for inference, it flips to 4-to-1; and for agentic AI, it reaches 1-to-1. This ratio shift means AMD's long-standing strength in data center CPUs—where it has been taking share from Intel—becomes critical. AMD projects this market will represent a $220 billion opportunity in the next few years. In Q2, data center CPU revenue surged 70%, and management guided for server CPUs to grow more than 80% year over year in the second half of 2025 and more than 70% in 2027.
AMD's GPUs have also gained traction in inference workloads. Unlike training, which prioritizes raw compute, inference depends on memory access and latency. AMD's chiplet design allows more high-bandwidth memory (HBM) to be packaged together, reducing latency—a critical advantage for inference performance. The company's most significant near-term catalyst is Helios, its first full-rack AI system, which integrates GPUs, CPUs, and networking components. AMD will begin deliveries this quarter, with shipments ramping in Q4. The company has secured major inference deals with OpenAI, Meta Platforms, Anthropic, and Microsoft, signaling strong customer demand.
For Q3, AMD guided revenue to grow 41% year over year to $13 billion (plus or minus $300 million), with adjusted gross margin expected at 56%. Management projects that the AI data center segment will see revenue more than double in 2027. While the stock is trading at a forward P/E ratio of 35 times 2027 analyst estimates, the forward PEG ratio is only 0.3, a metric below 1 that typically suggests undervaluation relative to growth. Despite the solid results and forward guidance, AMD shares fell, though the stock remains up around 125% on the year.
AMD's earnings mark a turning point in its positioning within the AI infrastructure market. While the company initially captured only limited share during the AI training phase—as hyperscalers relied primarily on Nvidia GPUs—the emerging focus on inference and agentic AI workloads creates a structural advantage for AMD. These downstream inference tasks depend heavily on CPU performance and memory efficiency rather than the massive parallel compute of training, fundamentally altering the hardware economics. AMD's data center CPU revenue surged 70% in Q2, and management now expects server CPUs to grow more than 70% in 2027, reflecting confidence in sustained demand. The announcement of Helios—combining GPUs, CPUs, and networking into an integrated system—signals that AMD is moving beyond component supply to offer complete systems for hyperscalers, a strategy that could drive both revenue scale and margin expansion. The company's adjusted gross margin rose to 56%, up 1,300 basis points year over year, driven by the higher-margin data center product mix. Management's projection of a $220 billion market opportunity in inference and agentic AI, coupled with confirmed deals at major cloud providers, provides visibility into demand. The stock's decline despite these results appears disconnected from the fundamentals, suggesting the dip may reflect broader market rotation or profit-taking rather than weakness in AMD's business trajectory.
AI-summarized, only the topics you pick — one digest a day via Email, Slack, or Discord.
Free · takes 30 seconds · unsubscribe anytime
Ask AI anything about this article. Q&As are published on this page for other readers too.
OpenAI completed a stock buyback worth roughly $7 billion, allowing current and former employees to sell share…

Brad Lightcap, OpenAI's Chief Operating Officer until April 2026 and a longtime executive who built the compan…

Brad Lightcap, who served as OpenAI's COO and most recently as special projects lead, announced his departure…

CEO Sundar Pichai announced that the Gemini app has surpassed 1 billion monthly active users, making it the 14…

Brad Lightcap, who joined OpenAI in 2018 and served as chief financial officer for four years before becoming…

River AI, founded by xAI co-founder Igor Babuschkin, raised $1.1 billion in a seed/Series A round led by Gener…

The AI news that matters, in one minute each morning.
Sign up free