ComputeX 2026 is about to open, with Jensen Huang, Lisa Su, and Charles Liang arriving in Taiwan to "secretly visit" the supply chain.

```

The three global AI chip giants are visiting Taiwan intensively on the eve of ComputeX 2026, reflecting that AI infrastructure construction is expanding from a GPU competition to cover the entire chain, including CPUs, ASICs, advanced packaging, and more.

According to a DIGITIMES report on Tuesday, industry sources revealed that Nvidia CEO Jensen Huang arrived in Taiwan on May 23 as the first to do so and is expected to stay for more than 10 days, with plans to visit TSMC founder Morris Chang, CEO C.C. Wei, and Quanta Chairman Barry Lam. AMD CEO Lisa Su visited Taiwan afterward. Intel CEO Pat Gelsinger will also arrive in Taiwan this weekend; in addition to internal meetings and discussions with TSMC executives, he has three secret meetings planned with the Taiwan supply chain and is scheduled to deliver a keynote speech at ComputeX on June 2.

The successive visits of the three major CEOs to Taiwan are seen by the industry as a key move to secure AI infrastructure capacity for the coming three to five years in advance.

Unexpected Surge in CPU Demand, AI Infrastructure Entering Full Value Chain Expansion

Six months to a year ago, almost no one in the market discussed CPU shortages, but this situation has quietly reversed. As AI inference truly takes off in the second half of 2025, CPU demand will explode rapidly, far exceeding previous market expectations. AMD predicts that the CPU market's annual compound growth rate (CAGR) will exceed 35% over the next five years.

Jensen Huang pointed out that the AI market has quickly extended from model training to the era of AI agents, inference, and AI factories. As AI begins to autonomously call various tools, a large number of CPUs are needed for data scheduling, system control, and memory management. Nvidia has officially entered the CPU market, its Vera CPU is now being sold independently, and it is optimistic about the future CPU market for AI data centers potentially reaching $200 billion. Huang also emphasized that GPUs will continue to be the core powerhouse of most AI computing.

On the Intel side, it was disclosed that in past AI training phases, the GPU-to-CPU configuration ratio was about 8:1, but it has now fallen to 4:1, and may approach or even reverse to 1:1 in the future. Intel estimates that in 2026, the global server CPU market and its own shipments will both maintain double-digit percentage growth, with this momentum continuing into 2027, and demand far exceeding supply.

Lisa Su's outlook on the AI industry is consistent with the above trends. Using baseball as an analogy, she stressed that the AI industry is still at a very early stage—"If AI were compared to a nine-inning baseball game, we’re only in the third inning now."

Intel’s Process Progress Key to Market Reassessment

For Pat Gelsinger’s visit to Taiwan this time, the outside world expects him to showcase a 'completely overhauled and rejuvenated Intel,' and engage in one-on-one exchanges with key supply chain partners to directly hear their opinions and suggestions.

In terms of process technology, Gelsinger recently admitted in an interview with CNBC that when he took over, the 18A process was not in ideal condition, but yields have now improved faster than expected. What has attracted even more market attention is the 14A process—its maturity, yields, and performance are currently ahead of where 18A was at the same stage of development, and design adoption and customer commitments are expected from the second half of 2026 to the first half of 2027.

Meanwhile, Intel is simultaneously boosting capacity and capital expenditures, focusing on EUV equipment installation and expanding advanced process capacity, with capacity set to increase each quarter.

In addition to technological progress, the moves of external customers for Intel’s foundry business are also being closely watched. Gelsinger revealed that more potential clients have proactively reached out to discuss using Intel's foundry services, and multiple external clients are scheduled to start production in the second half of the year.

On June 1, Intel plans to hold a supply chain cocktail party in Taiwan, inviting long-term upstream and downstream partners in Taiwan to attend; on the evening of June 2, a closed-door meeting is expected to take place with senior executives from the “five electronics giants,” ASUS, Advantech, etc., covering topics including AI servers, PCs, and U.S. manufacturing deployment and prospects.

Risk Warning and DisclaimerThe market has risks, and investment should be cautious. This article does not constitute personal investment advice, nor does it consider the specific investment objectives, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are suitable for their particular circumstances. Any investment based on this is at your own risk. ```