Monday, November 11th 2024
TSMC Can't Legally Make 2 nm Chips in the US Yet, Latest Nodes Must Remain in Taiwan
Even with billions of US dollars being invested overseas, TSMC cannot legally manufacture its most advanced nodes outside of Taiwan. According to Taiwan's Minister of Economic Affairs J.W. Kuo, "Since Taiwan has regulations to protect its own technologies, TSMC cannot produce 2-nanometer chips overseas currently." He added, "Although TSMC plans to make 2-nanometer chips [abroad] in the future, its core technology will stay in Taiwan." This provides crucial insight into TSMC's strategic positioning, both in its US expansion plans and in navigating global geopolitical waters, especially with Taiwan being the major hub of silicon innovation. Taiwan's semiconductor industry follows strict regulations regarding overseas production capabilities, requiring companies to maintain their most advanced manufacturing processes within Taiwan.
The company's international expansion strategy includes significant developments in the United States. TSMC's Arizona facilities are central to these plans, with multiple fabs in different stages of development. The initial Arizona facility will begin producing 4 nm chips imminently, while a second facility, scheduled to open in 2028, will manufacture then mature 3 nm and 2 nm chips. A third planned facility aims to produce 2 nm or more sophisticated chips. Meanwhile, Taiwan-based facilities will produce more advanced chips at the same time, with volume production of A-16 chips planned for late 2026, following the rollout of 2 nm chip production in 2025. Furthermore, Taiwan-US semiconductor cooperation will continue regardless of political changes. Taiwan Semiconductor Industry Association (TSIA) Chairman and TSMC Senior Vice President Cliff Hou noted that historical evidence suggests US electoral outcomes have not significantly impacted this technological partnership, though some adjustments may occur.
This situation raises critical questions about the effectiveness of the CHIPS and Science Act's objectives. Despite TSMC being awarded substantial US government support—including $6.6 billion in direct grants and up to $5 billion in loans for its Phoenix facilities expansion—Taiwan's legal restrictions on exporting leading-edge technology create a significant policy contradiction. The company cannot legally manufacture its most advanced chips on US soil, which could prompt concerns among US policymakers who have committed billions of taxpayer dollars to establish cutting-edge semiconductor manufacturing capabilities domestically. This disconnect between Taiwan's policies and US technological ambitions might lead to broader discussions about the return on investment for American taxpayers. While TSMC's Arizona fabs will indeed bring advanced manufacturing capabilities to US soil, they won't represent the absolute cutting edge of semiconductor technology.
Source:
Taipei Times
The company's international expansion strategy includes significant developments in the United States. TSMC's Arizona facilities are central to these plans, with multiple fabs in different stages of development. The initial Arizona facility will begin producing 4 nm chips imminently, while a second facility, scheduled to open in 2028, will manufacture then mature 3 nm and 2 nm chips. A third planned facility aims to produce 2 nm or more sophisticated chips. Meanwhile, Taiwan-based facilities will produce more advanced chips at the same time, with volume production of A-16 chips planned for late 2026, following the rollout of 2 nm chip production in 2025. Furthermore, Taiwan-US semiconductor cooperation will continue regardless of political changes. Taiwan Semiconductor Industry Association (TSIA) Chairman and TSMC Senior Vice President Cliff Hou noted that historical evidence suggests US electoral outcomes have not significantly impacted this technological partnership, though some adjustments may occur.
44 Comments on TSMC Can't Legally Make 2 nm Chips in the US Yet, Latest Nodes Must Remain in Taiwan
Until the US government does not do exactly that, the "CHIPS and Science Act" will be of little use and the US will continue to be riskily dependent on TSMC's factories in Taiwan.
And of course, this giant GlobalFoundries factory on US soil would also have to encompass not only the lithography process, but also all the subsequent steps necessary until the chips are ready for use by OEMs and domestic customers.
And a gigantic factory like this on American soil would generate millions of direct and indirect jobs, in addition to being able to provide and receive support to/from US research centers.
When we are talking their absolute latest and greatest it's lines with bad yields and processes that need time in the oven. While not having that kinda sucks...I would infinitely prefer a line of one or two generation old technologies that can pump chips out consistently, and that's basically required given how long mil-spec testing takes, than the best of the best that cost an arm and a leg to only be single digit percentages better than the far more mature processes.
I don't want to touch on the politics, but Taiwan having the costliest and least forgiving processes with months of wait to have their lithography machines delivered isn't a terrible thing. Heck, if we get stuck on another node for 3 generations the "best" technology out there will have processors costing a small fortune to have incremental improvements...where a significantly less expensive (and slightly lower performance) will be pennies on the dollar for costing and allow for much better cumulative performance with all of the multiple-CCD and multi-processor configurations on a single package.
If Chyna invades Taiwan, those TSMC factories aren't going to be making AMD stuff and NV stuff, they'll be expected to crank out any microchips required for the production of US Weaponry.
The fact that there is a "peace dividend" of domestic commercial chip manufacturing is the cherry on top, it's not the point.
Now I wonder, does Texas Instruments still has fabs?Given how they have shown over and over that they cant do anything right (without lying or bribing), I would be apprehensive about giving them more billions just to be squandered away.
Leading-edge nodes should re-appear after 12FDX launches. Since, GlobalFoundries is also in the 10nm/7nm FDSOI Europe alliance.
As well as Malta finally getting FDX production:
$1.5B from USA and €7.4B(both ST/GF) from EU. Means GlobalFoundries must [continue investing in the development of the next generation of FD-SOI technologies].
SOIL, WP2, T2.3 "Electrostatic and mobility enhancement" [(1) specific SOI substrates(sSOI) able to generate global constraints on transistors NMOS device couple with study to relax the strain locally for PMOS devices. (CEA; SOITEC)] Means that GlobalFoundries 7nm-target for Malta should be sooner than later. Global NFET strain with local PFET strain implies that GlobalFoundries won. The sSOI wafer will be tensile Si and not SiGe. Europe/USA there is no barrier for 7nm-class FDX at Dresden or Malta.
Intel has not received anything. Intel is being spread to thin to keep up with TSMC which is likely one of the reasons for their lackluster products.
Chip manufacturing is ridiculously expensive. TSMC spends 30-40 billion annually on R and D and plants which Intel can't keep up with. Intel has to spend the money first before they get any money back from the Chips act. Considering Intel not that long ago was stuck at 14nm and is almost caught up to TSMC for their highest performing node, it's clear where a huge portion of their spending has gone.
AMD fans just want to see Intel burn to the ground.