TSMC is not interested in buying Intel's fabs

TSMC Lobby
(Image credit: Taiwan Semiconductor Manufacturing Co., Ltd.)

Intel's struggles to land customers for its Intel Foundry unit raised concerns over viability of Pat Gelsinger's IDM 2.0 strategy and resurrected rumors about the company's possible plans to spin off or sell its manufacturing capacity. But only a few companies worldwide can potentially afford and operate Intel's fabs. One of them is TSMC, and it says it is not interested in acquiring those facilities. 

"Are we interested to acquire one of IDM's fabs?" C.C. Wei, chief executive and chairman of TSMC asked rhetorically during the company's earnings call when asked whether TSMC was interested in taking over Intel's fabs. "The answer is no, OK? No, not at all." 

Acquisition of Intel's fabs is a big deal for any company as taking over Intel's manufacturing capacity is challenging from financial, technological, regulatory, and perhaps even ideological standpoints. 

First up, Intel has vast manufacturing capacities. The company can challenge TSMC and Samsung Foundry in terms of output and cost tons of money to operate. Although TSMC has a lot of money (and can probably borrow), it would likely be hard for the company to finance a transaction of that size. 

Secondly, these fabs are used primarily to make Intel's own products on its own process technologies and are configured appropriately. While Intel's 18A and Intel 16 process technologies are compatible with industry-standard electronic design automation (EDA) tools and flows, many of Intel's other production nodes use the company's proprietary EDA tools and flows and are designed to offer high performance for Intel CPUs. To that end, the adoption of nodes like 14nm, Intel 7 (10nm ESF), Intel 4, and Intel 3 by third parties is complicated. Since Intel's fabs are already configured for these manufacturing technologies, it will be close to impossible (or at least extremely challenging technologically) to install additional tools and start making chips on other fabrication processes.  

Re-equipping these fabs will cost tens of billions of dollars, which would be a massive financial burden for any company. Yet, re-equipping fabs that produce its own products will be business as usual for Intel.  

Thirdly, should TSMC buy Intel's production assets, overlapping facilities and major differences would be abundant, resulting in additional costs and lower margins for the world's largest contract maker of chips. 

And with its $16.5 billion R&D budget, Intel spends loads of money on fundamental research beyond its current products and semiconductor production. While TSMC also conducts research, it does not spend as much as Intel on fundamental research projects, which is an ideological difference between the two companies. Also, Intel is known for large and expensive projects in general, whereas TSMC is known for its lean corporate structure. 

Finally, approving a possible acquisition of Intel's manufacturing assets by TSMC with authorities would likely be a nightmare. On the one hand, this will make TSMC the world's largest maker of chips without comparable rivals either in terms of production capacity or revenue (even now TSMC calls itself a Foundry 2.0 in an attempt to avoid antitrust scrutiny), which would not please regulators. On the other hand, it's highly unlikely the American government would allow a Taiwanese company to acquire Intel, one of the most prominent American companies for both geopolitical and political reasons, let alone technological ones. 

Anton Shilov
Contributing Writer

Anton Shilov is a contributing writer at Tom’s Hardware. Over the past couple of decades, he has covered everything from CPUs and GPUs to supercomputers and from modern process technologies and latest fab tools to high-tech industry trends.

  • Kamen Rider Blade
    Intel should stop "OutSourcing" their chip designs to TSMC.

    Build every part in-house.

    Then they'll have less to worry about with Fabs sitting still, not making anything.

    Doesn't matter if they can beat AMD or nVIDIA with the latest TSMC Production Node.

    Fab sitting Idle = Lost Revenue.
    Reply
  • hotaru251
    Kamen Rider Blade said:
    Fab sitting Idle = Lost Revenue.
    they are making some of their own stuff there (they make more than just the cpu) but even if they didnt I doubt they have the capacity to fill their demand physically.
    Reply
  • Kamen Rider Blade
    hotaru251 said:
    they are making some of their own stuff there (they make more than just the cpu) but even if they didnt I doubt they have the capacity to fill their demand physically.
    That's a Intel Foundary issue.

    They should've planned for this ahead of time.

    Intel shouldn't be dependent on ANY outside foundary for making their parts.
    Reply
  • Eximo
    They are making test chips on 18A now, so not like it is idle. Just not making production runs as of yet. Supposed to start making production runs in 2025. Do you want them to start early and have bad yields?

    I'm sure they looked at cost against getting a product to market. And some of Arrowlake is made by Intel, just not the GPU and logic chips.

    Not to mention Intel is still making their own PCH chips, as well as older CPU generations (they haven't stopped production of Raptor Lake yet), Xeon chips made on Intel 3 and 7, network chips, and the dozens of other products segments they are in.
    Reply
  • plutarchheavensbee62
    Well fancy that because Intel isn't interested in selling their fabs. Tomorrow the author should write "im not interested in selling my kidneys". Where do you find these writers. Toms is tantamount to the Onion with its absurdities.
    Reply
  • vinay2070
    Kamen Rider Blade said:
    Intel should stop "OutSourcing" their chip designs to TSMC.

    Build every part in-house.

    Then they'll have less to worry about with Fabs sitting still, not making anything.

    Doesn't matter if they can beat AMD or nVIDIA with the latest TSMC Production Node.

    Fab sitting Idle = Lost Revenue.
    They did try, but thier 20A did not work, and look at what happened when they pushed 13 and 14th gen CPUs on the older node? Though they just announced that they are sending samples of 18A chips to 3rd party, if it works, they wont use TSMC as much. The GPUs will still goto TSMC as they can be more power hungry.
    Reply
  • gg83
    So if Intel isn't number 1, it has to be sold? Companies struggle. Look at AMD since it's founding.
    Reply
  • wwenze1
    gg83 said:
    So if Intel isn't number 1, it has to be sold? Companies struggle. Look at AMD since it's founding.
    AMD sold their fab.

    Doesn't mean Intel has to, but... (*Points to Broadwell delay into nonexistence, points to 20 different 14nm nodes, points to 10ESF which finally arrived but did not bring the required improvements needed to catch up to N7, points to AMD laughing and selling Rembrandt-R Zen3 N6 cores in 2024 because of lack of competition*)
    Reply
  • svengollie
    Foundry is fancy name for job shop/CM, brittle business
    Reply
  • ezst036
    Kamen Rider Blade said:
    Fab sitting Idle = Lost Revenue.

    No doubt, however bad yields = lost revenue as well.
    Reply