Nvidia Corp CEO Jensen Huang (黃仁勳), who runs the semiconductor industry’s most valuable company, said the US is as much as 20 years away from breaking its dependence on overseas chipmaking.
Huang, speaking at the New York Times’ DealBook conference in New York on Wednesday, explained how his company’s products rely on myriad components that come from different parts of the world — not just Taiwan, where the most important elements are manufactured.
“We are somewhere between a decade and two decades away from supply chain independence,” he said. “It’s not a really practical thing for a decade or two.”
Photo: AFP
The outlook suggests there is a long road ahead for a key US President Joe Biden administration objective — bringing more of the chipmaking industry to US shores.
Biden has championed bipartisan legislation to support the building of manufacturing facilities in the US, and many of the biggest companies are planning to expand their US operations. That includes Taiwan Semiconductor Manufacturing Co (台積電), Nvidia’s top manufacturing partner, as well as Samsung Electronics Co and Intel Corp.
Europe is also pushing to build more manufacturing locally. It is part of an attempt to roll back decades of globalization that has scattered production across the world, but has also led to chokepoints in areas such as Taiwan and South Korea.
Separately, Huang reaffirmed his company’s commitment to China — still the largest market for chips. The company lost the ability to sell its most powerful artificial intelligence (AI) processors to the country after the US government imposed export restrictions, and then further tightened them last month. Washington believes that such steps are needed to protect national security.
Following the announcement of the latest rules, Nvidia is working on products for China that will not trigger restrictions, Huang said.
“We have to come up with new chips that comply with the regulation, and once we comply with the regulation, we’ll go back to China,” he said. “We try to do business with everybody we can. On the other hand, our national security matters. Our national competitiveness matters.”
He also cautioned against the unintended consequences of such rules. There are as many as 50 companies in China that are now working on technology that would compete with Nvidia’s offerings, he said.
Meanwhile, Huang said at the conference that artificial general intelligence (AGI) would be “fairly competitive” with humans in as few as five years.
When asked if AGI could be achieved in 10 years, Huang said it could happen even sooner.
“Depending on how you define it, I think the answer is yes,” Huang said. “If we define AGI as a piece of software — a computer that can take a whole bunch of tests and these tests reflect basic intelligence — and by completing those tests deliver results fairly competitive to a normal human, I would say that within the next five years, you can see AIs that can achieve those tests.”
LIMITED IMPACT: Investor confidence was likely sustained by its relatively small exposure to the Chinese market, as only less advanced chips are made in Nanjing Taiwan Semiconductor Manufacturing Co (TSMC, 台積電) saw its stock price close steady yesterday in a sign that the loss of the validated end user (VEU) status for its Nanjing, China, fab should have a mild impact on the world’s biggest contract chipmaker financially and technologically. Media reports about the waiver loss sent TSMC down 1.29 percent during the early trading session yesterday, but the stock soon regained strength and ended at NT$1,160, unchanged from Tuesday. Investors’ confidence in TSMC was likely built on its relatively small exposure to the Chinese market, as Chinese customers contributed about 9 percent to TSMC’s revenue last
With this year’s Semicon Taiwan trade show set to kick off on Wednesday, market attention has turned to the mass production of advanced packaging technologies and capacity expansion in Taiwan and the US. With traditional scaling reaching physical limits, heterogeneous integration and packaging technologies have emerged as key solutions. Surging demand for artificial intelligence (AI), high-performance computing (HPC) and high-bandwidth memory (HBM) chips has put technologies such as chip-on-wafer-on-substrate (CoWoS), integrated fan-out (InFO), system on integrated chips (SoIC), 3D IC and fan-out panel-level packaging (FOPLP) at the center of semiconductor innovation, making them a major focus at this year’s trade show, according
DEBUT: The trade show is to feature 17 national pavilions, a new high for the event, including from Canada, Costa Rica, Lithuania, Sweden and Vietnam for the first time The Semicon Taiwan trade show, which opens on Wednesday, is expected to see a new high in the number of exhibitors and visitors from around the world, said its organizer, SEMI, which has described the annual event as the “Olympics of the semiconductor industry.” SEMI, which represents companies in the electronics manufacturing and design supply chain, and touts the annual exhibition as the most influential semiconductor trade show in the world, said more than 1,200 enterprises from 56 countries are to showcase their innovations across more than 4,100 booths, and that the event could attract 100,000 visitors. This year’s event features 17
Hon Hai Precision Industry Co (鴻海精密), which assembles servers for Nvidia Corp, yesterday said that revenue last month rose 10.61 percent year-on-year, driven by strong growth in cloud and networking products amid continued front-loading orders for artificial intelligence (AI) server racks. Consolidated revenue expanded to NT$606.51 billion (US$19.81 billion) last month from NT$548.31 billion a year earlier, marking the highest ever in August, the company said in a statement. On a monthly basis, revenue was down 1.2 percent from NT$613.86 billion. Hon Hai, which is also a major iPhone assembler, added that its electronic components division saw significant revenue growth last month, boosted
RSS