Nvidia Corp’s Jensen Huang leads the parade of Taiwan’s AI computing leaders at Computex, Asia’s largest technology showcase. Over a jam-packed week, Keystones of Hardware in the Post-ChatGPT Era discusses the most pressing issues facing the industry, from growing bottlenecks in the supply of essential components like memory chips to the rise of challengers to Nvidia at the top of the semiconductor hierarchy.
Computex was originally a gathering of heavyweights from the PC era, but today it’s defined by conversations about the fundamentals of artificial intelligence competition. Some of the world’s biggest technology companies, from Nvidia to Intel Corp., are riding a wave of historic investments from companies like Meta Platforms Inc and OpenAI that are creating a new generation of billionaires and industry leaders. The gathering also comes amid growing concerns that the AI boom could enrich some companies and highly specialized workers while leaving others behind.
Here are some key themes, moments, and questions to keep in mind over the coming week.
The Rise of the Nvidia Challenger
Intel CEO Lip Vu Tan heads into town with the wind at his back. Tan, who has had a close relationship with President Donald Trump, has helped Intel rebuild credibility as a champion of U.S. semiconductor manufacturing, and the company’s stock price has increased sixfold over the past year.
Two other companies are looking to compete more directly with Nvidia. Qualcomm CEO Cristiano Amon will take the stage on Monday to talk about the company’s efforts to usher in the age of AI. According to Bloomberg News, Qualcomm has signed a deal with ByteDance to supply chips for data centers, a key victory for the company as it looks to expand from smartphone processors to AI infrastructure. TikTok founder ByteDance has invested US$70 billion (RM277.2) in capital expenditures this year, one of the highest spending by a company outside the US.
Arm Holdings CEO Rene Haas, a former Nvidia executive, will elaborate Tuesday on the company’s move to sell its own chips. Arm has been, and continues to be, the semiconductor blueprint maker for the entire industry, with companies like Qualcomm, Apple, and MediaTek all building on its designs. But the company is now starting to compete with Meta, counting it as its first major customer.
AI bottlenecks are growing
Just this week, a major Nvidia server maker warned that supplies are tight for a growing number of critical components, not just memory chips. WiWin Chairman Emily Hong said these restrictions may not be eased until late 2027. Before customers place an order, they ask if they have the parts or capacity they need.
“The prices of everything are going up at the same time,” Hong told Bloomberg News. “GPUs are expensive, and memory is also very expensive. We currently need additional funding to manage them.”
The most notable example is the surge in demand for Nvidia’s graphics processing units and memory chips, which the company expects to bring in US$500 billion (RM1.98 trillion) in annual revenue. IDC recalls that hyperscalers have shown a willingness to pay a premium to secure supply, with revenue expected to more than double to US$595 billion (RM2.36 trillion) this year.
Reflecting the soaring prices caused by these constraints, memory chip makers Micron Technology and SK Hynix this week joined Samsung Electronics in the US$1 trillion (RM3.96 trillion) market capitalization club. At the same time, the rapid growth in profits is causing dissatisfaction among workers who feel they are not getting enough of them.
It remains to be seen where the worst of the crisis will occur and whether it will significantly disrupt long and complex supply chains. Beyond memory, industry executives also talk about shortages in CPUs due to the proliferation of AI inference and agent platforms, and networking chips used to transmit data.
With Nvidia’s flagship Vera Rubin platform, optical interconnects are a hot new product, with expectations for increased demand for high-speed communications between processing nodes. This has accelerated the rally in a series of stocks previously unrelated to the AI infrastructure boom, including Nokia Oyge.
“AI infrastructure and physical AI are going to dominate the show,” said Anshel Sugg, principal analyst at Moor Insights & Strategy. “This is a combination of diversification and a desire to capture a piece of the fast-growing AI market.”
Chinese influence
One of the biggest undercurrents of the Taipei case will be what’s going on with its giant cross-strait neighbor, although few executives will raise the topic prominently, at least in public.
China’s future role in high-tech industry is one of the greatest uncertainties. China’s ambitions in AI and semiconductor manufacturing have alarmed Washington amid rising geopolitical tensions, especially with its island-based hardware sector under threat of attack.
Huawei created a stir this week by plotting a path to closing the gap with Taiwan Semiconductor Manufacturing Co., despite the U.S. government denying China access to cutting-edge semiconductor manufacturing equipment.
Executives are likely to raise questions about Huawei’s potential as a rival to both TSMC and Nvidia and its impact on Beijing’s campaign for technological independence.
How will PC manufacturers survive?
Most of the attention has been on training around AI, and as Lenovo Group and Dell Technologies expand as infrastructure providers, their stock prices have risen significantly. But Apple’s sudden entry into the budget laptop category has also put the traditional PC market in the spotlight. The US$599 (RM2,376) MacBook Neo was a success and raised expectations.
The Cupertino, Calif.-based company has stepped up the pressure just when margins on affordable devices are at their thinnest. According to data from inSpectrum, spot prices for certain memory products have increased more than 600% in the past year, leaving only large players like Apple and Lenovo able to control costs without raising prices on all of their products.
“The MacBook Neo is a huge thorn in the side of the PC industry,” said IDC Research Director Brian Marr. “We hope to see the PC ecosystem’s reaction to that during the show.” – Bloomberg
