Following a high-profile visit by Nvidia’s Jensen Huang, AMD CEO Lisa Su’s recent appearance at the AMD AI Developer Day in Shanghai signals a deepening commitment to the Greater China market. Addressing a crowd of developers and partners, Su presented a bullish timeline for artificial intelligence, forecasting that daily AI users will surge from 100 million in 2020 to over 5 billion by 2030. This projection underscores AMD's strategic pivot toward a world where AI is not just a feature, but a pervasive utility integrated into every facet of digital life.
AMD’s footprint in the region remains substantial despite ongoing geopolitical trade tensions. With over 4,000 engineers stationed across Beijing, Shanghai, Shenzhen, and Taipei, the company is positioning itself as a foundational partner for China’s digital infrastructure. Su noted that AMD’s EPYC processors already power over 700 cloud instances for China’s leading cloud service providers, supported by an ecosystem of over 100 software vendors and academic institutions. This local integration is critical as the industry shifts from experimental AI pilots to massive, enterprise-scale deployments.
A central theme of Su’s visit was the evolution of the 'AI Agent' and its implications for hardware architecture. In a dialogue with AI pioneer Lee Kai-fu, the discussion highlighted a 'tipping point' in AI's coding capabilities, moving from simple code assistance to end-to-end delivery of complex functions. This shift toward autonomous agents is driving a structural change in data centers; the traditional 1:4 ratio of CPUs to GPUs is rapidly evolving toward a 1:1 ratio. Su emphasized that as inference demands grow, the role of the CPU is becoming increasingly vital in supporting the full-stack computing power required for multi-agent collaboration.
For the Chinese market specifically, the constraints on hardware are breeding a unique form of innovation. Lee Kai-fu observed that because Chinese developers lack the luxury of 'brute force' computing due to limited high-end chip access, the ecosystem has pivoted toward extreme engineering efficiency. This focus on algorithmic optimization and lean infrastructure may ironically provide China with a competitive edge in the next phase of the AI race—the era of 'Autonomous Enterprises' where multi-layered agent networks drive ROI and respect data sovereignty.