AMD is pulling ahead in the AI accelerator market with strong Data Center revenue growth and confirmed large-scale customer deployments, while Intel is mid-turnaround, posting strong overall revenue but offsetting that with significant foundry losses and heavy capital spending.
AMD's business is winning today with higher-margin AI GPUs; Intel is asking for patience as it rebuilds its foundry operations and ramps newer CPU products.
What happened
AMD posted Q2 revenue of $11.536 billion, up 50.11% year over year, with Data Center revenue more than doubling to $6.718 billion and now representing 58% of total revenue. Intel's Q2 revenue reached $16.128 billion, up 25.42%, but a GAAP net loss of $11.033 billion (tied to a $12.53 billion CHIPS Act escrow charge) and Intel Foundry's $2.1 billion quarterly loss offset the top-line growth.
Why it matters
AMD's strength comes from premium-margin AI accelerators—Lisa Su confirmed Anthropic will deploy up to 2 gigawatts of MI450 series GPUs and stated that Helios delivers up to 15% more throughput at the same rack power and up to 30% more tokens per dollar than the competition. Intel is in the middle of a turnaround: Xeon 6 is ramping and DCAI grew 59%, but the company is spending heavily (2026 CapEx guided above $20 billion) on foundry manufacturing to catch up on next-generation chip production, which may appear as a drag on near-term profitability.
What to watch
AMD's Q3 guidance of approximately $13 billion and whether Instinct MI450 shipments meet Su's first-half 2027 Anthropic deployment timeline. For Intel, the key signals are 18A yield, EMIB-T ramp, and whether the company wins external foundry customers beyond the $293 million in quarterly external foundry revenue it currently reports. Su also told analysts that AMD's data center segment revenue should more than double year-over-year in 2027.
AMD and Intel each posted their strongest quarters in years, but their trajectories diverge sharply. AMD's Q2 revenue hit $11.536 billion, up 50.11% year over year, with Data Center revenue surging to $6.718 billion—more than doubling and now accounting for 58% of total revenue. CEO Lisa Su told investors that Anthropic will deploy up to 2 gigawatts of MI450 series GPUs, underscoring real commercial adoption at scale. Su also emphasized that Helios delivers up to 15% more throughput at the same rack power and up to 30% more tokens per dollar than the competition, signaling a performance and efficiency advantage that is resonating with customers. In a forward-looking statement, Su told analysts that AMD's data center segment revenue should more than double year-over-year in 2027.
Intel's Q2 narrative is noisier. Revenue reached $16.128 billion, up 25.42%—what CEO Lip-Bu Tan called the "strongest revenue growth in more than fifteen years." Xeon 6 is ramping fast, and the data center AI (DCAI) segment grew 59%, signaling momentum in CPU-based inference workloads. However, these gains are overshadowed by a GAAP net loss of $11.033 billion, which includes a $12.53 billion CHIPS Act escrow charge, plus a $2.1 billion quarterly loss from Intel Foundry Services. To address its manufacturing gap, Tan is committing heavy capital: 2026 CapEx is guided above $20 billion to catch up on next-generation node execution.
The strategic difference is stark. AMD operates as a design house, dominating dedicated AI accelerators and data center GPUs—claiming to be one to two full hardware cycles ahead in raw specifications and an order of magnitude ahead in commercial adoption and revenue. AMD's non-GAAP gross margin of 56% reflects the premium positioning of its accelerator business. Intel is pursuing vertical integration, combining renewed CPU strength (Xeon 6 and DCAI growth) with an in-house foundry effort designed to manufacture its own leading-edge chips and serve external customers. However, Intel's GAAP gross margin of 40.4% and its foundry losses reveal the near-term cost of this strategy.
The article frames the test ahead: for AMD, whether Instinct MI450 shipments hit Su's first-half 2027 Anthropic timeline and whether Q3 guidance of approximately $13 billion holds; for Intel, whether 18A yield and EMIB-T ramp improve, and whether the company wins external foundry customers beyond the current $293 million in quarterly external foundry revenue. Both stocks have cooled—AMD down 14.98% over the past month and Intel down 11.04%—suggesting market caution even as the earnings results showed strength. The article concludes that AMD offers higher-quality growth today, while Intel offers a turnaround story with more upside variance, backed by $30 billion in cash and short-term investments. The verdict: AMD is the business winning now; Intel is asking for patience.
AMD and Intel are pursuing fundamentally different strategies in the AI infrastructure race. AMD has positioned itself as a design house focused on high-margin accelerators and CPUs, riding the wave of dedicated AI GPU demand where the company claims to be one to two full hardware cycles ahead in pure specifications and an order of magnitude ahead in commercial adoption and revenue. Intel, by contrast, is betting on vertical integration—combining CPU strength (Xeon 6 ramping strongly, with DCAI growing 59%) with a rebuilding foundry business aimed at manufacturing next-generation chips in-house. The financial picture reflects these different trajectories: AMD's Q2 gross margin of 56% non-GAAP underscores the profitability of its GPU-centric model, while Intel's 40.4% GAAP gross margin and its $2.1 billion foundry loss in Q2 show the near-term cost of the turnaround. Both firms face distinct risks: AMD must manage export controls and defend against a gaming revenue decline, while Intel carries execution risk on foundry yield and manufacturing ramp, alongside the burden of spending 2026 CapEx above $20 billion to close the gap.
The article frames this as a race between design leadership and vertical integration. AMD's customer list—anchored by Anthropic's 2-gigawatt MI450 commitment and comparable wins across frontier AI labs—suggests the market is validating the pure-play accelerator model in the near term. Intel's counterargument relies on longer-term optionality: the company holds $30 billion in cash and short-term investments to fund its rebuild, and if 14A (its next-generation process node) wins external foundry customers and 18A achieves high yields, the integrated model could offer better cost and control down the line. For now, the article makes clear that AMD is "the business winning today," while Intel is "the business asking for patience."
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.
The AI news that matters, in one minute each morning.
Sign up free