🔥 As demand for AI semiconductors explodes globally, even TSMC—the world's largest foundry—can't keep up with production. With demand reportedly three times current capacity, could this be a once-in-a-generation opportunity for Japan's Rapidus? Here's the inside story of the global race for cutting-edge 2nm chips.
TSMC's Production Capacity Hits Its Limits
The semiconductor giant TSMC faces unprecedented supply pressure. Despite announcing record-breaking capital expenditure of approximately 9 trillion yen for 2026, the company cannot keep pace with surging demand for AI chips.
TSMC CEO C.C. Wei has acknowledged that production capacity remains "very tight." In an unusual move, NVIDIA's Jensen Huang personally visited Taiwan to request increased wafer supply. Industry estimates suggest AI chip demand has reached roughly three times current production capacity.
Why Expansion Can't Keep Up
Manufacturing AI semiconductors requires cutting-edge EUV lithography equipment and advanced packaging technology like CoWoS. However, deploying this equipment takes considerable time—production plans cannot translate instantly into increased output.
Furthermore, while TSMC is converting mature process facilities to advanced process and packaging operations, it still cannot meet demand. Although 2nm mass production began in late 2025, orders from major customers like NVIDIA, Apple, and AMD continue to flood in, making it extremely difficult for new customers to secure production slots.
A Business Opportunity for Rapidus
Amid this supply shortage, Japan's Rapidus is attracting significant attention. At its manufacturing facility "IIM" under construction in Chitose, Hokkaido, the pilot line began operations in April 2025. By July, the company successfully confirmed operation of 2nm GAA transistors.
Jim Keller, the legendary semiconductor engineer and Tenstorrent CEO, evaluated Rapidus's prototype data as "very good." Tenstorrent became Rapidus's first publicly announced customer.
Rapidus's Differentiation Strategy: Short TAT
Rapidus's key differentiator is "Short TAT (Turn Around Time)." While traditional foundries process hundreds of wafers in batches, Rapidus employs single-wafer processing. This approach promises to dramatically reduce cycle time from prototyping to mass production.
In the AI market, speed has become the competitive axis. As the industry shifts from general-purpose to application-specific chips, Rapidus's production methodology could deliver value distinct from TSMC's offering.
Challenges and Future Outlook
Rapidus targets 2nm mass production in the latter half of fiscal 2027—approximately two years behind TSMC. However, if TSMC cannot resolve its supply shortage during this period, Rapidus could have a prime opportunity for customer acquisition. Cumulative government support has reached approximately 2.9 trillion yen, and the company aims for IPO around fiscal 2031.
Nevertheless, significant challenges remain: improving yield rates, controlling manufacturing costs, and securing talent. Additionally, TSMC, Samsung, and Intel are all accelerating expansion investments. Depending on when supply shortages are resolved, Rapidus's advantages could diminish.
Significance for Japan's Economic Security
Rapidus's significance extends beyond mere business competitiveness. For Japan, which has depended on overseas manufacturing for cutting-edge semiconductors, having domestic production capability is crucial from an economic security perspective. With geopolitical risks intensifying, supply chain diversification has become a global trend.
TSMC's supply shortage could represent a "once-in-a-generation opportunity" for Rapidus. Whether this opportunity can be seized, however, depends on technological development and customer acquisition over the next two years.
In Japan, discussions continue with mixed expectations and concerns about Rapidus. How does your country view the AI semiconductor supply situation and the trend toward reshoring semiconductor manufacturing? Please share your thoughts in the comments!
References
Reactions in Japan
Is it really true that even TSMC can't meet demand? I really hope Rapidus succeeds. Show them Japan's technological prowess!
What happens if they fail after investing over 2 trillion yen of taxpayer money? It's the people's money!
The short TAT differentiation strategy is interesting. Not competing head-on with TSMC seems wise.
If Jim Keller approves, maybe there's hope. That guy really knows his stuff.
From an economic security perspective, having domestic capability for cutting-edge semiconductors is essential. We shouldn't judge solely on cost.
As a Hokkaido resident, I have high hopes for Rapidus. It'll create local jobs too.
Prototyping success doesn't guarantee yield rates in mass production. We shouldn't be too optimistic.
Even with TSMC's 9 trillion yen investment, it's not enough? Just how massive is AI demand?
Domestic semiconductor production isn't a pipe dream. Japan has excellent equipment and materials manufacturers.
Even if production starts in 2027, won't TSMC be even further ahead by then? Can we catch up?
AI chip demand three times supply? That's also an investment opportunity. Better check related stocks.
I haven't forgotten past failures of Japan's semiconductor ventures. Please succeed this time.
If a Taiwan contingency happens, we might lose access to TSMC. Domestic production is needed as insurance.
If they pay engineers properly, Japan still has talented people. I have expectations for Rapidus.
Amazing that CEO Huang went personally to request more wafers. Shows how serious the supply shortage is.
I'm based in the US. I've always thought our dependence on TSMC was too risky. If Japan's Rapidus succeeds, it will help diversify the supply chain. Rooting for them.
From Germany's perspective, Japan's challenge to domestically produce cutting-edge semiconductors is interesting. The EU has similar initiatives, but Japan seems to be progressing faster.
I work in India's IT sector. We're experiencing the surge in AI demand firsthand. Chips are in short supply globally. It makes sense for Japan to enter this market.
I'm a researcher in France. Rapidus's single-wafer processing approach is academically fascinating. If successful, it could represent a paradigm shift in semiconductor manufacturing.
From Korea here. Samsung is struggling with 2nm too, so I'm curious how far Rapidus can go. Honestly though, I think catching up to TSMC will be difficult.
I work at a UK tech company. Considering geopolitical risks, we want more chip procurement options. It would be great if Rapidus becomes a viable choice.
From Mexico. Semiconductor supply chain issues have greatly impacted the automotive industry. We welcome new supply sources.
I'm from Taiwan, living in the US. Knowing TSMC's capabilities, I understand how challenging Rapidus's endeavor is. But competition is good for the industry.
Based in UAE. AI investment is accelerating in the Middle East too. Stable semiconductor supply is an important issue for us as well. Watching Japan's efforts closely.
I'm an engineer from Poland. Japan's approach of challenging cutting-edge technology with government support is instructive for EU countries.
From Australia. As US-China tensions deepen, Japan having semiconductor production capability will contribute to stability in the Asia-Pacific region.
I work at a Brazilian startup. For startups like ours, difficulty procuring chips is a serious issue. We hope supply increases.
I'm a Russian-American tech investor. The 5 trillion yen investment is significant, but success could yield substantial returns in this market.
I'm a Japanese engineer based in Silicon Valley. Interest in Rapidus is growing here too. Trust in Japanese manufacturing technology is still strong.
I'm a financial analyst in Singapore. Rapidus's plan to IPO around 2031 is ambitious, but achievable depending on market conditions.