The elephants in the Computex room

Image credits: Computex

This year’s Computex, the first since Taiwan eased pandemic-related travel restrictions, was a celebration of the global computer and chip industry. But between the exhibits, speeches, and product announcements, like Jensen Huang’s Nvidia keynote (just one day before the company hit a trillion-dollar valuation), several topics were barely touched upon, at least publicly. The fact of the matter is that, between issues such as geopolitical tensions and AI-induced chip shortages, the semiconductor industry is in great turmoil. Here are some of the things left mostly unsaid at Computex.

1. Geopolitics makes life more difficult for chip makers

As relations between the US and Chinese governments continue to grow more frosty, things are getting tougher in the semiconductor industry. The ongoing war between the two countries over the chip industry will have a growing impact on how semiconductor supply chains are managed, especially for superchips required in generative AI and other high-power computing tasks .

Last October, the United States passed new export laws requiring US chip makers to obtain a license from the Commerce Department before exporting advanced chips, including those used in AI and chip-making equipment, to China. The United States, Japan and the Netherlands have also reached an agreement to stop exports of chip-making tools to China. Firms caught up in the new restrictions included Nvidia, which had been barred from selling A100 and H100 GPUs in China, costing up to $400 million. Both chips are used for training large language models such as OpenAIs GPT-4. In response, Nvidia made a slower chip for sale in China.

In turn, China has opened an investigation into US memory chip maker Micron over cybersecurity concerns before banning some chips from being sold. The ban could have benefited Micron’s competitors in China such as Samsung Electronics and SK Hynix, but the United States has reportedly asked South Korea not to plug the Chinese market gap if Micron is banned. It all happened in the last half year, and it’s no exaggeration to expect the US-China coup de grace to have a dramatic impact on the global semiconductor industry in the coming months.

The world’s largest chipmaker with 59% global market share, TSMC, is based in Taiwan and while it may not face the same fines as American companies, many of its biggest customers are in China. As a result, TSMC reportedly hedged its bets. TSMC has suspended production of advanced chips for Chinese startup Biren to ensure it complies with US regulations, according to a Bloomberg report released shortly after the US enacted the sanctions.

TSMC, like all other Taiwanese semiconductor companies, is also in charge of Taiwan-China relations. TSMC has built smelters in the US and Japan, but most of its manufacturing is still in Taiwan, which leaves open questions about what will happen to its chips, which much of the world’s technology companies rely on, if Taiwan reports -China continue to intensify. .

2. How TSMC’s work culture will translate

TSMC plans to spend $40 billion on its two Arizona factories, which produce advanced chips. TSMC founder Morris Chang said Taiwan’s work culture is one of the reasons it powers the best semiconductor companies in the world. For example, he said TSMC’s 24-hour on-call practices mean that if a piece of equipment fails at 1 a.m., it will be fixed within the hour, as opposed to 9 a.m. in the United States. But the intensity of TSMC’s labor practices has come under scrutiny, including in a report earlier this month by The New York Times that found that it and other companies with a similar work culture face high turnover, despite the prestige to work for them.

With the opening of TSMC’s first foundry in Arizona, it may also be the road to a culture clash. The EE Times interviewed a chief engineer in the US who said, more or less, that the culture needs to change, but the working environment and general requirements have been established in Taiwan for a long time. So this will be transplanted to TSMC Arizona. A few small changes should make it more acceptable, but American engineers will have to adapt to the working environment and this type of culture.

3. Talent shortage

Employee attrition and lack of talent in general have the potential to be a major headache for semiconductor companies around the world as industry growth is expected to outpace the increase in skilled workers. In a recent report, Deloitte estimated that more than a million additional workers will be needed globally by 2030, or more than 100,000 a year. In the United States, there are fewer than 100,000 graduate students enrolled in electrical and computer engineering, and the US semiconductor industry may soon face a shortage of some 70,000 to 90,000 workers. Unless something changes, that means CHIPS Acts’ attempt to turn the US into a semiconductor powerhouse may simply lack the necessary manpower.

4. Shortage of AI chips

Human talent isn’t the only thing in short supply. Generative AI works on chips, mostly GPUs made by Nvidia, but these are becoming increasingly scarce. Microsoft is reportedly facing an internal shortage of the server hardware it needs to run its AI, and according to the WSJ, OpenAI CEO Sam Altman said during a May 16 congressional hearing that it would be better if less people use ChatGPT due to processor bottleneck. . Some server makers and direct customers told the WSJ that they’re waiting more than 6 months to get Nvidia’s latest GPUs. DigiTimes reported earlier this month that Nvidia has placed multiple orders for chips requiring TSMC’s wafer-on-substrate (CoWoS) chip packaging in an effort to ease the bottleneck. The chip shortage could cause massive amounts of stress for AI companies and AI startups, but it’s one reason Nvidia stock has soared to a trillion-dollar valuation.

Meanwhile, startups and big companies like Intel and NTT are working on alternatives like photonic chips. As my colleague Kyle Wiggers reports, photonic chips use light to send signals instead of electricity like conventional processors. In theory, this means higher training performance because light produces less heat than electricity, can travel faster, and is less susceptible to temperature changes and electromagnetic fields. But there are several catches. For one thing, photonic chips are larger and harder to mass-produce, and their architectures still rely on electronic control bottlenecks, which can create bottlenecks. Second, they require a lot of power to convert the data into a format that the chips can work with. Finally, signal regeneration, or the process of regenerating degraded optical signals during transmission through photonic chips, means that signals can become distorted over time. As a result, it may take years for photonics technology to go mainstream (even with photonics AI startups like Lightmatter getting large amounts of funding), and in the meantime, generative AI companies will keep scrambling for GPUs. .

It is important to note, however, that there are rumors that the AI ​​bubble could burst at some point, especially if regulators become more cautious and start taking action. The chip industry overproduced during the pandemic in response to shortages, and as a result, there is now a surplus of memory chips.

5. Taiwan’s drought impacts wood chip supplies

With rainy weather in Taipei last week, it might be hard to believe for out-of-town Computex visitors, but Taiwan is experiencing another drought. The previous one in 2021 negatively impacted the country’s semiconductor manufacturing because chip manufacturing requires a huge amount of water. TSMC, for example, uses more than 150,000 tons a day. During the last drought, it relied on truckloads of water to keep producing chips.

This time, TSMC is prepared, not only with leased water tanks but also with new wells. He told Nikkei Asia that he has contingency plans for different stages of water restriction and works with government and private organizations to save water and develop water resources. It has also implemented water conservation measures at its facilities in Southern Taiwan Science Park, including reducing water consumption and recycling wastewater.