🏭 What if Japan could make the same cutting-edge chips that power ChatGPT and the latest iPhones — right in its own backyard?
That's exactly what's about to happen. TSMC, the world's largest chipmaker, just announced it will produce 3-nanometer semiconductors in Kumamoto, Japan by 2028 — chips so advanced that only Taiwan makes them today. The investment? A staggering $17 billion. Here's why this is a game-changer for Japan and the global tech landscape.
From 7nm to 3nm: A Dramatic Upgrade
On February 5, 2026, TSMC Chairman and CEO C.C. Wei made a personal visit to Japan's Prime Minister Sanae Takaichi to deliver some big news: TSMC is completely overhauling its plans for its second factory in Kumamoto.
The original plan was to produce 7-nanometer chips — already respectable technology used in many smartphones and computing devices. But TSMC has decided to skip ahead and go straight to 3-nanometer production, one of the most advanced chipmaking technologies currently in mass production anywhere in the world.
To put it simply, the "nanometer" number refers to how small the circuits on a chip are. The smaller the number, the more powerful and energy-efficient the chip. Today's most advanced AI processors from companies like NVIDIA, and the latest chips in Apple's iPhones, are built using 3nm technology. Until now, these chips could only be made in Taiwan.
The total investment for TSMC's Kumamoto operations is now expected to reach approximately $17 billion (about ¥2.6 trillion), up significantly from the original $12.2 billion plan for less advanced chips.
Why the Sudden Change?
Several forces converged to make this upgrade happen.
The AI boom changed everything. When TSMC first planned its Kumamoto expansion, the automotive semiconductor market was the main target. But the explosive growth of artificial intelligence — from ChatGPT to autonomous driving systems — has created enormous demand for the most advanced chips. According to Omdia analyst Akira Minamikawa, the 7nm chips originally planned for Kumamoto are becoming a shrinking market, while AI-related demand is rapidly shifting to 3nm and even more advanced technologies.
Geopolitical risk is accelerating diversification. The potential for conflict in the Taiwan Strait has made governments and companies worldwide nervous about relying on Taiwan for virtually all cutting-edge chip production. Japan is positioning itself as a critical backup manufacturing base. As PM Takaichi noted, having 3nm production capability in Japan carries significant meaning from an "economic security" perspective.
Japan's generous subsidies sealed the deal. The Japanese government has already committed up to ¥732 billion (roughly $4.8 billion) in subsidies for TSMC's second Kumamoto plant, and is reportedly considering additional support for the upgraded plan. The Ministry of Economy, Trade and Industry plans to quadruple its budget for advanced semiconductor and AI support to approximately ¥1.23 trillion in the next fiscal year.
The Kumamoto Story So Far
TSMC's presence in Kumamoto has been building steadily. The first factory, operated through subsidiary JASM (Japan Advanced Semiconductor Manufacturing), began mass production in late 2024, producing chips using more mature 12nm to 28nm technology. That factory, a joint venture with Sony, Denso, and Toyota, already employs about 2,400 people and has achieved manufacturing performance on par with TSMC's facilities in Taiwan.
The economic impact has been significant. The Kyushu region, already nicknamed "Silicon Island" for its concentration of semiconductor companies, has seen an estimated ¥6.9 trillion ($45 billion) in economic ripple effects projected over ten years from TSMC-related investments. About 90 companies have established or expanded operations in the Kumamoto area since TSMC announced its entry.
However, the journey hasn't been entirely smooth. Construction of the second factory was delayed due to traffic congestion issues caused by the first factory's operations. There were also concerns that the automotive chip market — the original target — wasn't growing as fast as expected, raising questions about whether the factory would run at full capacity.
The pivot to 3nm production effectively answers those concerns by targeting the red-hot AI chip market instead.
What Will These Chips Be Used For?
Industry experts caution that Kumamoto's 3nm factory won't necessarily be producing the absolute cutting-edge AI training chips used by companies like NVIDIA — those will likely continue to be made in Taiwan and potentially at TSMC's Arizona facility in the United States.
Instead, Kumamoto's 3nm production could serve several critical markets. AI inference chips used in data centers to run trained AI models are one likely application. Edge AI processors for autonomous vehicles and robotics — areas where Japanese companies like Toyota have enormous demand — are another strong candidate. Advanced processors for next-generation smartphones and computing devices could also be produced.
The key insight from several analysts is that having any 3nm production capability in Japan dramatically changes the equation for Japanese companies. Equipment makers and materials suppliers will benefit from having a cutting-edge factory nearby, potentially sparking a broader renaissance in Japan's semiconductor ecosystem.
Japan's Bigger Semiconductor Comeback
TSMC's Kumamoto upgrade is just one piece of Japan's ambitious semiconductor revival strategy.
In Hokkaido, domestic startup Rapidus is building a factory aimed at producing even more advanced 2nm chips, with backing from IBM's technology and approximately ¥2 trillion in government support. The Japanese government has reportedly concluded that Rapidus and TSMC will serve different market segments rather than competing directly.
The national goal is bold: triple the combined revenue of domestic semiconductor manufacturers to ¥15 trillion by 2030, compared to 2020 levels. Between TSMC's investments in Kumamoto and Rapidus's ambitions in Hokkaido, Japan is essentially building two separate paths back to the frontier of chip manufacturing — one through foreign partnership, and one through domestic innovation.
Challenges Ahead
Not everyone is celebrating without reservation. Several important questions remain.
The first is about customers. Advanced chip manufacturing is enormously expensive, and factories need committed, high-volume customers to justify the investment. Japan currently lacks major domestic companies designing chips at the 3nm level. Whether Japanese AI startups and traditional manufacturers can create enough demand remains unclear.
The second concern is workforce. The semiconductor industry worldwide faces severe talent shortages, and Japan is no exception. JASM plans to grow its Kumamoto workforce to over 3,400, but finding and training engineers capable of running 3nm production lines is a significant challenge.
Third, there's the question of whether this is truly a "Japan" success story or primarily a Taiwanese company operating on Japanese soil. As one industry observer noted, if Kumamoto remains simply a production site, Japan gains jobs and tax revenue but limited technology transfer. The real prize would be accumulating knowledge in process improvement, yield optimization, and equipment development — capabilities that could lift Japan's entire semiconductor ecosystem.
Finally, the plan is still in early stages. Sources familiar with the matter emphasize that details are still being discussed and could change. The semiconductor industry moves fast, and what seems like the right bet today might look different in two years.
A Pivotal Moment
Despite the caveats, TSMC's decision to bring 3nm production to Japan represents a watershed moment. Just five years ago, the idea that cutting-edge chips would be manufactured in Japan seemed like a distant dream. The country had steadily lost ground in semiconductor manufacturing since the 1990s, watching as Taiwan, South Korea, and the United States pulled ahead.
Now, with $17 billion in investment, strong government backing, and the tailwind of insatiable AI demand, Japan is positioning itself as a serious player in the advanced chip manufacturing game once again.
What about your country? Is your government investing in semiconductor manufacturing? Do you think having domestic chip production matters for economic security, or is the global supply chain working fine as is? Share your thoughts — we'd love to hear perspectives from around the world.
References
- https://www.bloomberg.com/jp/news/articles/2026-02-05/T9YJDOKK3NY800
- https://www.nikkei.com/article/DGXZQOGM295DG0Z20C26A1000000/
- https://www.trendforce.com/news/2026/02/05/news-tsmc-reportedly-to-upgrade-kumamoto-2nd-plant-from-67nm-to-3nm-boosting-japans-chip-capabilities/
- https://semiwiki.com/semiconductor-manufacturers/tsmc/363007-tsmc-kumamoto-pioneering-japans-semiconductor-revival/
- https://newspicks.com/news/15980530/
Reactions in Japan
As a local, I'm genuinely happy about this. But honestly, how many people in Kumamoto can actually work at a 3nm fab? Three of my colleagues left for JASM, but they're all on the 12nm line. I worry whether workforce training can keep up with this pace.
Let's be real — TSMC switched to 3nm because 7nm is no longer profitable for them. It looks like they're doing Japan a favor, but it's really just a business decision. Why build a 7nm line in Kumamoto when their Taiwan 7nm lines are already underutilized?
Behind this accelerated 3nm timeline is a calculation to elevate Japan as a second strategic base for next-gen logic chips. The goal is to seize initiative in supply chain restructuring post US-China decoupling. But the looming risk of a Taiwan contingency remains a wild card.
We're pouring over $6.5 billion in taxpayer money into this, and in the end isn't it just TSMC's factory that profits? The crucial question is how much technology transfer actually happens, and I feel like that discussion is severely lacking.
Semiconductor stocks are going crazy today. Tokyo Electron and SCREEN are near limit-up. This could be the start of Act 2 for equipment and materials stocks. Installing 3nm production lines means more demand for EUV lithography tools — the question is how much of that ASML-driven demand Japanese equipment makers can capture.
Rapidus targets 2nm by 2027, TSMC targets 3nm by 2028. And TSMC has a proven track record. Isn't Rapidus's reason for existing getting thinner by the day? I'm worried this is becoming a double investment of taxpayer money.
The news is all rosy, but as someone living in Kikuyo town, the thought of even worse traffic congestion is terrifying. Rent and prices have gone up unbelievably over the past two years. I wish they wouldn't forget the perspective of people actually living here.
What they'll actually manufacture here is key. AI chips like GPUs are physically large, making yield a challenge at cutting-edge nodes. Which customer and product they start with during the 3nm ramp-up will define the factory's character. Smartphone SoCs like Apple's might be the realistic starting point.
Kyushu's GDP is currently about $330 billion. SMBC Nikko Securities estimates it could grow to $500 billion by 2035. If 3nm mass production materializes, even that estimate might look conservative. The question is whether Kyushu's infrastructure development can keep pace with investment.
As an analog circuit designer, 3nm honestly feels distant from my world. But having TSMC's cutting-edge line in Japan means process development knowledge could flow to equipment and materials makers. The indirect benefits might be significant.
This news should be interpreted cautiously. TSMC's 7nm utilization rates in Taiwan aren't great, and by 2028, 3nm won't be cutting-edge anymore. In semiconductor manufacturing where utilization rates are everything, an announcement when demand forecasts aren't solidified requires careful scrutiny.
PM Takaichi's 'win-win partnership' line is diplomatic, but having TSMC's CEO personally visit the PM's office is extremely unusual. It shows how seriously TSMC values the Japanese market. With Trump tariffs raising US investment costs, this is proof that Japan's relative attractiveness is increasing.
This news is incredible timing right before job hunting season. JASM internship competition is going to be insane, but having the chance to work on world-leading semiconductors while staying in my hometown was unimaginable just a few years ago. Maybe the trend of seniors all leaving for Tokyo will finally change.
Nobody is squarely addressing the water and power issue. A 3nm fab consumes even more water and electricity than 7nm. Is Kumamoto's groundwater supply okay? This directly connects to the nuclear power restart debate, but politicians keep avoiding that topic.
It's bittersweet that Japan, which dominated the world with DRAM in the 80s, is now celebrating the arrival of a foreign company's factory. But realistically, Japan doesn't have the capability to build 3nm on its own right now. Choosing practicality over pride is the right call.
I work at TSMC's Arizona fab, and honestly, internally everyone's talking about how Japan's operations are running smoother. We're plagued by labor shortages and union issues causing delays. If Japan starts 3nm mass production first, the pressure on our fab is going to ramp up even more.
As a Taiwanese person, I have mixed feelings. TSMC technology going overseas raises security concerns. But considering Taiwan contingency risks, diversifying production makes sense. I want Japan to succeed, but I hope core technology stays in Taiwan.
For Samsung in Korea, this is nothing but a threat. If Japan becomes a 3nm production hub, the Japan-Korea power balance in equipment and materials shifts even more in Japan's favor. Our government needs to get way more serious about semiconductor support or we'll be left behind.
India has just started pouring money into semiconductor manufacturing, but 3nm is honestly a dream within a dream for us. Japan can pull this off because of its industrial accumulation. India is still scrambling to secure a 28nm fab.
TSMC is building a fab in Dresden too, but we're still at the 28nm level. If Japan jumps to 3nm, Europe falls even further behind. The EU Chips Act budget is nowhere near sufficient.
As a supply chain researcher, what I'm watching is how much Japan can actually become a 'maker.' Having a TSMC factory doesn't change much if there's no design capability. Developing Japanese fabless companies is the next challenge.
From China's perspective, the Japan-US semiconductor alliance is part of the containment strategy. But honestly, the decision to produce 3nm domestically is sound. SMIC is working hard here too, but the restriction on EUV equipment access is devastating.
As a Silicon Valley startup CEO, having a fab in Japan means faster design iteration, which is great. When we depended only on Taiwan, convincing investors about geopolitical risk was a nightmare.
I work at a semiconductor packaging plant in Vietnam. If Japan starts 3nm mass production, demand for back-end processes could increase across ASEAN. Looking forward to the ripple effect across the entire supply chain, not just front-end.
Working in Mexico's auto industry, Japan's 3nm production signals acceleration of autonomous driving tech. Toyota's investment in JASM makes total sense. Level 3+ autonomous driving absolutely requires these advanced chips.
As an Australian energy analyst, what concerns me is power consumption. A 3nm fab's electricity demand is massive. Japan risks its AI and semiconductor strategy being derailed by power shortages unless it accelerates nuclear restarts.
Penang in Malaysia is growing as a semiconductor hub too, but advanced processes are still a distant dream. Japan's materials and equipment base is why TSMC sees investment value there. What we need first is educational infrastructure for talent development.
From Russia's perspective, this is pure envy. Due to sanctions, we can't even get mature process chips, let alone cutting-edge ones. Countries like Japan that can partner with allies are fortunate. The winners and losers of geopolitics are becoming crystal clear.
As a French industrial policy planner, Japan's 'subsidies plus foreign investment attraction' model is instructive. The EU is focused on supporting STMicroelectronics, but we need to face the reality that countries that attracted TSMC are reaping the biggest benefits.
I work at Intel's Ireland fab, and watching Intel's recent struggles, I can't help but think TSMC's foundry model was the right answer all along. Japan was smart not to try building everything domestically like Intel, and instead partnering with TSMC.