Advanced Micro Devices, Inc. (AMD) Earnings

Management highlights

### General Business & Strategic Positioning - AMD delivered its sixth consecutive quarter of over 30% year-over-year revenue growth, with record revenue and profitability, and is tracking materially ahead of the long-term financial model shared at November 2025 Financial Analyst Day. - Management expects overall high-performance AI computing market to grow ~40% annually through 2030, reaching ~$2 trillion by 2030, and AMD expects to grow well above this market rate. - AMD now expects to substantially exceed its prior long-term target of over 35% annual revenue growth, and to significantly beat its $20 annual EPS target within the strategic timeframe. ### Data Center Business Updates - Epic server CPUs delivered the fifth consecutive quarter of record revenue, with cloud and enterprise sales each growing over 70% year-over-year, exceeding prior quarter guidance. AMD gained x86 server revenue share year-over-year, with the 6th Gen EPYC Venice family (built on 2nm Zen 6 technology) now in production, with major OEMs set to launch platforms and cloud providers starting deployments in late 2026. Customer demand for Venice is stronger than any prior EPYC generation. - Data center AI accelerator revenue more than doubled year-over-year, driven by strong demand for Instinct MI350 series accelerators. AMD launched the Helios rack-scale AI platform combining EPYC Venice CPUs, MI450 series GPUs, Pensando networking, and Radeon software, which outperforms competitors on inferencing throughput and cost efficiency. - Helios customer demand is ahead of initial forecasts. AMD announced new large-scale strategic partnerships: Anthropic will deploy up to 2 gigawatts of MI450 series GPUs starting H1 2027, and Microsoft will deploy Helios at scale on Azure for frontier model inferencing, adding to existing multi-gigawatt deployments with OpenAI and Meta. Helios initial shipments begin Q3 2026, ramping through Q4 and into 2027. - The AMD Radeon software stack for AI reached a key inflection, with over 3 million models running out-of-the-box on AMD Instinct, and open source contributions increased more than tenfold year-over-year. The new Rackham.ai AI-assisted development platform delivers over 2x training performance and 3x inferencing performance over prior versions. ### Client, Gaming and Embedded Business Updates - Client business grew 23% year-over-year, driven by record mobile revenue and continued commercial share gains, with Ryzen Pro sales growing over 50% year-over-year. AMD launched the next-generation Ryzen AI Halo developer platform capable of running 300-billion-parameter models locally, with a partnership to include one year of Hugging Face Pro with every system. - Gaming revenue declined 31% year-over-year due to lower semi-custom console sales at this stage of the console cycle, and lower gaming graphics revenue due to higher component costs suppressing demand. - Embedded segment delivered its strongest year-over-year growth in over three years, with broad-based demand across networking, aerospace and defense, and communications end markets. AMD is tracking to another record year of new design wins, totaling over $18 billion led by major wins in core embedded end markets.

Guidance

- Q3 2026 total revenue is expected to be approximately $13 billion, plus or minus $300 million, representing 41% year-over-year growth and ~13% sequential growth. This guidance expects strong double-digit sequential growth for both the Data Center and Embedded segments, with a modest sequential decline in the Client and Gaming segment (slight client growth offset by a double-digit decline in gaming). - Q3 2026 non-GAAP gross margin is expected to be approximately 56%, non-GAAP operating expenses are expected to be ~$3.65 billion. - For H2 2026, server CPU revenue is expected to grow more than 80% year-over-year. - For full year 2027, server CPU revenue is expected to grow more than 70% year-over-year (off a significantly higher 2026 base). - For full year 2027, total Data Center segment revenue is expected to more than double year-over-year, with upside potential to exceed this baseline target. - AMD projects the data center AI accelerator market will grow more than 45% annually to ~$1.4 trillion by 2030, and the server CPU market will grow more than 50% annually to ~$220 billion by 2030, with agentic AI server infrastructure becoming the largest and fastest growing segment of the server CPU market by 2030. - Operating expense growth is expected to be lower than top-line revenue growth over the next several years to drive operating leverage and earnings growth.

Segment performance

Total company Q2 2026 revenue was a record $11.5 billion, up 50% year-over-year and 13% sequentially. Total operating income was $3.1 billion, representing a 27% operating margin. 1. Data Center Segment: Revenue hit a record $6.7 billion, which more than doubled year-over-year and increased 16% sequentially. This segment contributed 58% of total company revenue. Operating income for the segment was $2.1 billion, equal to a 31% operating margin. Growth was driven by a more than 70% year-over-year increase in Epic server CPU sales (with record enterprise sell-through and robust cloud demand) and Instinct AI accelerator sales that also more than doubled year-over-year. 2. Client and Gaming Segment: Total segment revenue was $3.8 billion, up 6% year-over-year and 7% sequentially, contributing 33% of total revenue. Client business revenue was $3.1 billion, up 23% year-over-year and 6% sequentially, led by record mobile processor revenue. Gaming business revenue was $779 million, down 31% year-over-year primarily due to lower semi-custom console sales, though it grew 8% sequentially. Total segment operating income was $582 million, equal to a 15% operating margin. 3. Embedded Segment: Revenue was $977 million, up 19% year-over-year and 12% sequentially, contributing 8.5% of total revenue. Operating income was $386 million, equal to a 40% operating margin, up from 33% year-over-year, driven by higher revenue and favorable product mix.

Risks & headwinds

- Higher industry-wide memory and component costs are expected to weigh on overall PC and gaming graphics demand in H2 2026, and may impact overall PC market performance in 2027. - The server CPU supply chain was tight in H1 2026 due to unforecasted demand growth, though management expects supply conditions to improve in 2027. - Ramping new 2nm process technology for new product generations is inherently challenging, and requires coordinated expansion of end-to-end supply chain capacity including wafers, back-end processing, packaging, and substrates. - Large-scale AI data center build-out timelines and customer capacity to accept new hardware could impact the pace of Helios ramp in 2027, though management does not currently see constraints that would impact baseline guidance targets. - Data center AI accelerator gross margins are expected to be slightly below AMD's corporate average, creating mix headwinds that must be offset by gross margin expansion in higher-margin segments like server CPUs and embedded.

Analyst Q&A

• Q: When will data center GPU revenue cross over server CPU revenue, and which business will drive more 2027 growth? What are the gross margin implications of this mix shift? /

  A: Management expects very strong double-digit growth for both server CPUs and data center AI GPUs through H2 2026 and into 2027, with both businesses serving as major growth drivers. The data center AI TAM is larger, so substantial AI growth is expected as Helios ramps in 2027. Server CPU growth is accretive to overall gross margins, while data center AI has a slightly lower gross margin than the corporate average; the growing embedded segment also provides gross margin tailwinds, and management expects to maintain and improve overall gross margins through the mix shift.

• Q: Which of data center CPU or GPU will grow more sequentially from Q2 to Q3, and can AMD outgrow the 40% overall market CAGR through 2030? /

  A: Both server CPUs and data center AI GPUs will deliver strong double-digit sequential growth in Q3, with additional growth in Q4: server CPU supply increases throughout the second half, and Helios AI ramps meaningfully starting in Q4. All AMD business segments benefit from AI tailwinds, and management confirmed AMD expects to grow faster than the overall 40% market CAGR through 2030, gaining share across all segments.

• Q: What is the 2027 gigawatt deployment line of sight, and how does higher HBM memory content impact margins and supply? /

  A: AMD confirmed overall data center revenue will more than double in 2027, with strong ramping of existing large strategic customer deployments, and sufficient supply to meet guidance with upside potential. Revenue per gigawatt is expected to land in the double-digit billion dollar range, as currently forecasted. Management has secured sufficient HBM memory allocation for 2027 planned shipments; the larger memory footprint is a performance advantage for large models, but AMD can adjust memory configurations for smaller workloads to meet customer TCO requirements, and remains on track to hit profitability targets.

• Q: Is server CPU supply currently constrained, and can the supply chain support over 70% 2027 growth? /

  A: Server CPU supply was tight in H1 2026 due to faster-than-forecast demand growth, but management has expanded end-to-end supply chain capacity over the past quarters, and expects 2027 supply conditions to be better than 2026 enough to support over 70% year-over-year growth, with upside potential if demand remains stronger than expected. AMD's chiplet design reduces the volume of new 2nm wafers needed for ramp, easing new node supply constraints, and the company has secured all necessary component capacity to meet planned growth.

• Q: What is the customer diversity for the early MI450/Helios ramp, and are data center infrastructure constraints a risk? /

  A: In addition to the announced large gigawatt-scale frontier model customers (OpenAI, Meta, Anthropic), there is strong demand from many smaller-scale customers, creating good customer diversity starting in Q4 2026 and H1 2027. Management confirmed no visible infrastructure constraints that would impact baseline guidance targets; the entire ecosystem is motivated to bring on AI capacity faster, and there is upside potential if build-outs accelerate faster than currently planned.