Q1 2024 Summary
Published Jan 10, 2025, 5:10 PM UTC- AMD is well-positioned to capture growth opportunities in large-scale data center systems, especially in AI compute, and believes they are "very well positioned to play in those large-scale systems."
- The client business is performing well, with strong demand for Ryzen processors, and AMD expects client revenue to be up sequentially in the second quarter, with opportunities to gain additional commercial PC share based on performance and efficiency advantages.
- Demand for AMD's data center GPUs is "really strong," with customers moving from initial POCs to pilots to full-scale production, building backlog and engagements, indicating strong future growth.
- MI300 profitability is currently below corporate average and will take time to improve. Jean Hu stated that MI300 is a complex product in its first year of ramp with ongoing yield and process improvements, and over time, gross margin should be accretive to corporate average.
- Data center operating profit declined 19% sequentially despite 2% revenue growth, due to increased investments and lower GPU margins. Jean Hu explained that increased investments in AI and lower gross margins from GPUs were major drivers for the decline in operating income.
- Potential competitive pressure in server CPUs from a competitor launching high-core-count products, which could impact pricing and market share. Lisa Su acknowledged the competitor's product launch but stated that while pricing is stable, increased competition could pose challenges.
-
MI300 Outlook and Supply Constraints
Q: What's driving the $500 million increase in MI300 guidance, and how are you handling supply constraints?
A: The MI300 ramp is going well, with increased visibility from both current and new customers committing to MI300, allowing us to raise the full-year outlook from $3.5 billion to $4 billion. Supply is tight in the near term, but we've successfully ramped supply—the fastest product ramp we've done—and will continue to ramp every quarter this year. -
MI300 Demand vs Supply Constraints
Q: Is the $4 billion MI300 outlook supply constrained or demand constrained?
A: The $4 billion guidance is not supply capped; we have supply capability above that. In the near term, such as in Q2, demand exceeds supply, but overall we have good visibility and are working to increase supply to meet strong AI demand. -
Server CPU Growth Expectations
Q: How much can server CPU grow this year, and how back-end loaded will it be?
A: Our data center segment is on a very strong ramp in the second half, including both server CPUs and data center GPUs. While we won't provide specific numbers, we expect overall segment growth to be very strong double digits. -
Data Center Margins and GPU Impact
Q: Are data center margins affected by GPU profitability, and when will GPUs become accretive?
A: Data center GPU gross margins are currently below the data center average due to initial ramp costs and increased investment in AI. Over time, GPU gross margins will be accretive to the corporate average, but it will take a while to reach server CPU levels. -
Server CPU Competitive Pricing
Q: Is there risk of aggressive pricing impacting your server CPU ASPs?
A: Our server CPU average selling prices are stable. The pricing environment is stable, and our performance and energy efficiency translate into total cost of ownership advantages for our customers. -
Embedded Business Outlook and Gross Margins
Q: How does a slower rebound in embedded impact gross margin expansion?
A: The embedded business declined more than expected due to weaker demand but is expected to gradually recover in the second half. The primary driver for gross margin expansion is the strong performance of the data center business, which will continue to be the major driver. -
MI300 Profitability and Margins
Q: When will MI300 profitability exceed the corporate average?
A: While we're still in the first year of ramping MI300, a complex product, we expect that over time its gross margin should be accretive to the corporate average as process improvements continue. -
Turin Server CPU Transition
Q: How quickly will the Turin server CPU ramp within your portfolio?
A: Turin uses the same platform as Genoa, making for an easier ramp. Turin will coexist with Genoa for some time, but we expect a faster transition than from Milan to Genoa, expanding our addressable market with Turin. -
Factors Driving Faster MI300 Ramp
Q: Is the faster MI300 ramp due to customer collaboration or aggressive build-out plans?
A: It's both; we're confident in broadening the ramp as we've optimized key models with customers and passed important milestones. There's significant demand for AI compute, and we're working closely with customers to meet that demand. -
Rack-Scale Systems and System Integration
Q: Do you have all the pieces needed for rack-scale systems in your roadmap?
A: System integration is critical, and we have all the pieces, including our Infinity Fabric, Pensando's networking expertise, and partnerships with key networking companies. We're specifying full system reference designs for future generations. -
Participation in Mega Data Centers
Q: Can AMD participate in mega data centers, or is it winner-takes-all?
A: We believe it's not winner-takes-all and are well-positioned to play in large-scale systems. Customers desire optionality in building out large training environments, and we view this as a very attractive area for AMD. -
MI300 Customer Demand
Q: Has there been any change in MI300 customer demand or intentions?
A: Demand remains very strong, with customers moving from initial proofs of concept to pilots and full-scale production. We feel good about ongoing deployments and are building backlog and engagements. -
Data Center GPU Roadmap
Q: How does your data center GPU roadmap beyond MI300 compare to competitors?
A: We have a multi-year, multi-generational roadmap beyond MI300, informed by close collaboration with top AI customers. Our chiplet architecture allows flexibility, and we're confident we'll continue to be highly competitive. -
MI300 Supply Capacity
Q: Do you still have supply capacity beyond the increased MI300 guidance?
A: Yes, we have supply visibility significantly beyond the updated $4 billion guidance, ensuring we can meet demand that exceeds current expectations. -
Server CPU Performance vs Competitors
Q: What drove your server CPU performance compared to competitors?
A: Our server business performed well, with strong double-digit year-over-year growth. Strength in enterprise offset some seasonality, and we continue to gain market share. -
Client PC Outlook and AI PCs
Q: What's the outlook for your client PC business and AI PC impact?
A: We expect client revenue to be up sequentially in Q2, with increases in units and ASPs in the second half. Our AI PC products are suited for premium segments, with broader adoption across the portfolio in 2025. -
Ultra Ethernet Consortium and Competitiveness
Q: How important is the Ultra Ethernet consortium for your system competitiveness?
A: We're supportive of the open ecosystem and the Ultra Ethernet consortium, but it's not a limiter to our ability to build large-scale systems. We're working internally and with partners to enable Ethernet-based networking solutions. -
Gaming Weakness Impact
Q: How is significant gaming weakness impacting your overall guidance?
A: Gaming is down more than 30% sequentially in Q2 due to weak demand and high inventory levels, with the second half expected to be lower than the first. However, gaming has lower gross margins than our corporate average, so reduced gaming revenue positively impacts our overall gross margin mix.