TSMC CEO C.C. Wei says, ‘It will be a long time before we can meet customer demand’ — tells shareholders that he will keep prices stable, refrain from implementing price hikes

News
By
Published

All this demand for AI chips is just too much for one company.

TSMC Logo
(Image credit: Getty Images / Bloomberg)

TSMC CEO C.C. Wei has told company shareholders that it still won’t be able to completely address the production demands for AI chips in the years to come. Even though the company has opened multiple fabs across the world, including the one in Arizona, the insatiable demand for the most advanced processors means that there still isn’t enough production capacity to go around for all customers. Bloomberg reports that the additional capacity that is expected to go online in the U.S. is still not enough to feed the increasing demand from hyperscalers.

“It will be a long time before we can meet customer demand,” C.C. Wei said. Despite that, TSMC is still forecasting a 30% increase in sales this year. It’s also not taking advantage of the supply bottleneck, with Wei adding that the company will avoid sudden price hikes similar to the memory and storage chip market industry’s experience to ensure business stability.

Hyperscaler buildouts are expected to hit $725 billion just this year. And unless the AI bubble bursts, demand is only expected to go up every year. TSMC has been building many new fabs in Taiwan, the U.S., and other parts of the world. But building these manufacturing plants would take years, and it seems that semiconductor manufacturing demand would outpace supply if the current AI infrastructure build-out continues.

Latest Videos From

TSMC Arizona’s manufacturing capacity has been sold out through 2027 since early 2025, showing the massive demand for the company’s output. The company is continually expanding this site, too, with its Taiwan headquarters authorizing a $20 billion capital injection last month to continue the development of Fab 21 phase 2. It’s expected that this would allow the company to start mass producing 3nm chips in Arizona in 2027, which is about a year earlier than the original 2028 launch date. There have also been reports of additional fabs and other units, bringing the total Arizona site to 12 fabs, 4 packaging facilities, and an R&D center.

These production shortages mean that TSMC can expect that their expansion plans will have customers once they’re completed. However, this is also an opportunity for Intel, which is trying to win customers for its 18A and 14A processes. Both Apple and Nvidia are reportedly considering Intel for some of their 2028 chip production, and sources say that the former has already reached a preliminary agreement with Team Blue.

The lack of availability has also led Elon Musk into semiconductor manufacturing. Even though building chips is a totally different beast when compared to building electric cars and even rocket ships, it seems that the billionaire founder is ready to take on the challenge with Terafab. It seems that he’s pretty serious, too, with his team already in talks with various suppliers and that they’re willing to pay a premium to ensure priority.

Google Preferred Source

Follow Tom's Hardware on Google News, or add us as a preferred source, to get our latest news, analysis, & reviews in your feeds.

TOPICS
Jowi Morales
Jowi Morales
Contributing Writer

Jowi Morales is a tech enthusiast with years of experience working in the industry. He’s been writing with several tech publications since 2021, where he’s been interested in tech hardware and consumer electronics.

21 Comments Comment from the forums
  • bit_user
    Money is too cheap. Raise interest rates. That will help curtail investment in AI, which will slow down purchasing and free up some capacity.

    That's actually in line with standard orthodoxy: increase interest rates to quell inflation. We're seeing inflation ticking up, partly fueled by the AI boom. So the natural response should be to increase interest rates. However, that doesn't seem likely to happen, any time soon.
    Reply
  • trica
    We should only be worried about one number - the amount of energy each data center goes through, measured at the main power line coming in from the street. Who cares what an individual search costs? I want figures that include ALL power usage, whether it's from a customer query, running the cooling fans, the lights on the building, the AC in the boss's office or the microwave in the break room. Those electrons all get spent because the data center exists, and don't care what particular device they were powering at the time.
    Reply
  • SkyBill40
    bit_user said:
    Money is too cheap. Raise interest rates. That will help curtail investment in AI, which will slow down purchasing and free up some capacity.

    That's actually in line with standard orthodoxy: increase interest rates to quell inflation. We're seeing inflation ticking up, partly fueled by the AI boom. So the natural response should be to increase interest rates. However, that doesn't seem likely to happen, any time soon.

    Ahh, but the present administration doesn't take heed of that mantra. And I'll leave it at that so as to not get hit with the hammer.
    Reply
  • sseemaku
    bit_user said:
    Money is too cheap. Raise interest rates. That will help curtail investment in AI, which will slow down purchasing and free up some capacity.

    That's actually in line with standard orthodoxy: increase interest rates to quell inflation. We're seeing inflation ticking up, partly fueled by the AI boom. So the natural response should be to increase interest rates. However, that doesn't seem likely to happen, any time soon.
    Reality will be, the companies will squeeze other sectors and direct all the available money to AI.
    Reply
  • Hooda Thunkett
    It will be a long time before we can meet customer demand...
    Nah, one recession could bring demand in line with production. That doesn't take much time at all.
    Reply
  • Zaranthos
    Don't worry, Terafab will replace the production you weren't able to provide once they make all the chips they need for satellites, cars, and Tesla bots. :tonguewink:

    This is only half a joke. Musk built Colossus in record time and now rents it to the competition.
    Reply
  • bit_user
    Zaranthos said:
    Don't worry, Terafab will replace the production you weren't able to provide once they make all the chips they need for satellites, cars, and Tesla bots. :tonguewink:

    This is only half a joke. Musk built Colossus in record time and now rents it to the competition.
    The example of Colossus is actually more instructive than you might've thought. The reason why he's renting it to Anthropic for inference is because it was constructed so hastily that it's actually really bad for training, which was its real goal. So, renting it out is just a way to keep it from being a near total loss, financially.

    So, "bravo Elon!" Style over substance.

    P.S. He repurposed an old abandoned factory, thereby having ready infrastructure in place and avoiding the need for a lot of the build that most datacenters have to do. That's great when you can find such sites, but there are only so many.
    Reply
  • Zaranthos
    bit_user said:
    The example of Colossus is actually more instructive than you might've thought. The reason why he's renting it to Anthropic for inference is because it was constructed so hastily that it's actually really bad for training, which was its real goal. So, renting it out is just a way to keep it from being a near total loss, financially.

    So, "bravo Elon!" Style over substance.

    P.S. He repurposed an old abandoned factory, thereby having ready infrastructure in place and avoiding the need for a lot of the build that most datacenters have to do. That's great when you can find such sites, but there are only so many.

    But talk about when life gives you lemons make lemonade. Colossus 2 will be paid for by the rental revenue of Colossus 1 which will also be fully paid off in record time. Google is also renting some of the compute and long term compute contracts are in effect. It also wasn't useless for training Grok, just inefficient. The media does a good job of painting it as a failure when it was ultimately a resounding success and financially outperformed most other data center projects by a lot.
    Reply
  • bit_user
    Zaranthos said:
    But talk about when life gives you lemons make lemonade.
    It's not "life" that gave him lemons. He gave them to himself by rushing into the project and making such a big deal about doing it quickly that he did it badly. That was a foreseeable and avoidable failure. All it would've taken is just bringing in enough expertise to guide the design of the datacenter, and then not being totally reckless on the timescale.

    If he really believed he needed it to be done that quickly, then he also needed it not to be done badly. Doing it badly totally undermines any benefit he gained by doing it quickly.

    "Run fast and break things" is about software, which is generally a lot easier to fix than hardware & infrastructure. Given Musk's experience with Tesla and SpaceX, he should understand the difference, and appreciate taking the necessary time on hardware.

    Zaranthos said:
    Colossus 2 will be paid for by the rental revenue of Colossus 1
    No, I doubt that. I think he probably won't even recoup even the full investment on Colossus 1, before the hardware becomes too obsolete.

    Zaranthos said:
    It also wasn't useless for training Grok, just inefficient.
    I never said it was useless. But, the reason he funded it with debt is because AI is a huge race. So, even being inefficient is a killer, because they scaled it to meet their needs and if it doesn't, that means they're falling behind.

    Zaranthos said:
    The media does a good job of painting it as a failure
    I don't know about that, but you sure do provide Mr. Musk some free PR.
    Reply
  • Zaranthos
    bit_user said:
    I don't know about that, but you sure do provide Mr. Musk some free PR.
    Pretty sure my ramblings are essentially worthless PR for most anyone. :grin:

    The revenue potential for Colossus 1 is around 70 billion based on contracts that already exist. Mistakes, shortcomings, and failures may have existed for Colossus 1 but based on cold hard facts it still may very well be wildly profitable. Bypass the media and read the SEC filings about the inked contracts.
    Reply