Chen Liwu: AI computing power shortage may extend until 2029

Intel CEO Chen Liwu attended the "DBS Asia Foresight Summit" today.

[Reporter Wu Xintian/Singapore Report] Intel CEO Chen Liwu and Broadcom President and CEO Chen Fuyang shared the stage with DBS Group Executive President Chen Shushan at the "DBS Asia Foresight Summit" to discuss in depth the latest developments in the semiconductor industry and the structural changes brought about by AI. Chen Liwu mentioned that the shortage of computing power is expected to continue until 2029.

At the beginning, Chen Shushan specifically mentioned the deep friendship between classmates and work between Chen Liwu and Chen Fuyang over the years. From the early years when the two lived and worked together, they have now reached the point where they are now in charge of global technology giants. Chen Liwu recalled that in the face of geopolitics, supply chain fluctuations, and regulatory challenges in various countries, trust among industry leaders, compliant operations, and continuous innovation have always been the core for enterprises to move forward steadily.

Talking about the development of artificial intelligence that is of most concern in the current market, Chen Liwu pointed out that AI has a profound impact on improving daily life and work efficiency, and its long-term transformative power is expected to even surpass the past Internet revolution. However, due to the complexity of introducing AI in various industries and the exponential demand for computing resources, computing power facilities will continue to face tight supply, and this shortage is expected to continue until at least 2029. In addition, in addition to chip and memory bottlenecks, power shortages and rising capital costs will also become important constraints on industrial development. High-end capital costs are more likely to become the main bottleneck for subsequent AI expansion.

In terms of technology development and value transfer, Chen Fuyang analyzed that in addition to general-purpose graphics processors, customized accelerators tailored for specific large-scale language model workloads will play a key role in improving computing efficiency and cost-effectiveness in the future. Although hardware and chip manufacturers are currently benefiting greatly from the construction boom, the core long-term value of AI is shifting to upstream applications and cutting-edge models. At the same time, Asia has occupied a favorable position in the overall ecosystem by virtue of its solid advantages in semiconductor manufacturing and hardware supply chains, and it also has significant growth potential in the field of AI applications in the future.

Regarding the sustainability of AI spending, the two said that the long-term rationality of large-scale capital expenditures by enterprises ultimately depends on the actual return on investment. As generative AI demonstrates concrete business benefits in automating processes, predicting capabilities, and improving productivity, the market will strike a balance between driving innovation and managing security risks. The two CEOs finally emphasized that in the face of rapidly evolving technological waves and high capital costs, leaders must remain humble and make wise decisions to push the industry toward a more sustainable and positive impact.