AMD CEO Lisa Su on AI Compute, Data Centers, and Future Tech at CES

AMD's AI Compute Strategy

• 💡 Lisa Su highlights AMD's new MI455 rack-scale system solution, featuring the world's first 2nm chip, designed to meet the escalating demand for AI compute.
• 🚀 The company anticipates a massive increase in AI users, projecting growth from 1 billion to over 5 billion active users in the next five years, necessitating significant compute power.
• 📈 AMD is addressing diverse AI use cases with a family of chips, including the MI440 for enterprise applications, enabling upgrades in existing data centers.
• 🎯 The focus is on delivering outstanding performance with an advantageous total cost of ownership and championing an open ecosystem and deep partnerships.

Addressing Compute Deficits and Bottlenecks

• ⚠️ Su acknowledges a substantial compute deficit, estimating a need for a 100x increase in compute power over the next four to five years, potentially reaching yottaflop scale.
• 🧩 The industry faces multiple constraints, including silicon production, memory supply, and power availability, requiring a coordinated ecosystem approach.
• 🤝 AMD emphasizes partnership across the supply chain to ensure resources are available for expanding the compute environment.

Future of Computing and Market Strategy

• 🚀 AMD aims to provide turnkey solutions that are easy for customers to deploy, optimizing full systems for AI compute.
• ⚡ Future generations like MI500 are projected to offer a thousand times the performance of the MI300 generation through advanced engineering and technology.
• 🇨🇳 China remains an important market for AMD, with ongoing efforts to secure licenses for shipping advanced chips, despite regulatory complexities.
• 📊 AMD is targeting tens of billions of dollars in AI revenue by 2027, signaling a significant inflection point in technology and partnerships.

PC Market and AI Integration

• 💻 AMD has gained market share in the PC market through its product portfolio and early bet on AI PCs.
• 💡 AI PCs are expected to drive upgrade cycles, with AMD also developing new AI development systems.
• ⚠️ Supply chain constraints, particularly on the memory side, are being monitored due to high demand from AI data center compute.

Physical AI and Robotics

• 🤖 AMD is actively involved in physical AI, leveraging its expertise in FPGAs and embedded systems, powering numerous robotic applications.
• 🤝 Partnerships, such as with Generative Bionics, are advancing humanoid robot capabilities and intelligence.
• ⚙️ The business model involves providing components for real-time local capabilities and the underlying technology for training and inference on humanoids.

AI's Economic Impact and Future Outlook

• 📈 AI is demonstrably improving company productivity, enabling faster product development and enhanced business processes.
• 🌐 The aggregated impact of AI is expected to influence the global GDP over the coming years.
• ✨ Su expresses optimism for 2026, stating, "you ain't seen nothing yet," predicting continued rapid progress in AI models, use cases, and tangible business outcomes.