Intel CEO Lip-Bu Tan said at a conference held by DBS in Singapore that, dragged down by continued constraints in the technology supply chain, artificial intelligence hardware will remain in tight supply for the next two to three years. At the Singapore event on October 12, he systematically laid out core trends such as AI and quantum; and when he issued his shortage verdict, memory chips and other core components were persistently scarce, with prices rising month after month, and TSMC's packaging capacity had hit the ceiling. As an executive who keeps a close eye on chip processes and new materials, Lip-Bu Tan has always been sensitive to the direction of the industry.
Agents lift the CPU, quantum within three years
What is even more intriguing is the shift in demand structure. Lip-Bu Tan pointed out that the spread of agentic AI is fueling demand for CPUs rather than GPUs, which has also become his lever for reviving Intel: the company's second-quarter Data Center and AI segment revenue soared 59% year over year, while the old rule of thumb in AI training of "8 GPUs paired with 1 CPU" has been pulled down by agents to a 4 to 1 ratio. He also mentioned that token costs of China's frontier models are significantly lower, and combined with such low prices, their performance is already satisfactory. In the long run, capital is the primary constraint on expansion, but disproving doubts about return on investment is the key to stabilizing capital confidence; spending money to build infrastructure is only the first half of the game, and the second half will be about who can deliver real, tangible returns.
On quantum computing, Lip-Bu Tan is optimistic, expecting it to go mainstream in the next three to five years, and reiterated that the core of technology adoption lies in finding the right landing scenarios—a judgment that echoes the key investments disclosed through his venture fund at the end of last month.