
What happened
Daiwa House Industry says it can industrialize data centers with prefab techniques from its homebuilding past, building units about the size of a convenience store that skip the wait for extra-high-voltage power.
Why it matters
If prefabricated units can be built without that grid connection, data-center capacity could be added faster in places where securing power is the bottleneck, according to the company's approach.
WHO IT HITSCompanies that need to place AI servers and data-center developers in Japan face power-connection waits that can exceed a decade in some regions, making prefab units an option worth evaluating for faster deployment.
Summaries like this, every morning.
Daiwa House Industry's proposal comes as GPU servers for generative AI and machine learning draw roughly five times more power per rack than conventional servers, according to the company. That jump in power demand collides with a grid that, in some regions, cannot deliver extra-high-voltage power for more than a decade after an application. The result is a bottleneck that affects both companies that need data-center space and the developers that build it.
The scale gap is stark: Japan has 242 data-center facilities against 4,165 in the U.S., according to the Ministry of Internal Affairs and Communications' Information and Communications White Paper. The comparison is not apples-to-apples, but Japan has fewer data centers relative to its population and economic size.
On top of power constraints, the construction industry faces labor shortages, rising material costs, and longer build times tied to work-style reform. Daiwa House Industry's answer is to bring prefabrication—the technique it has refined since its founding as a housing maker—to data-center construction. Units about the size of a convenience store can be built without waiting for extra-high-voltage power, a shift the company frames as industrializing data-center development.
Pick your industry and the AI tools you use, and get news related to your work every day.
Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →
Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.
Berkshire's $363 billion stock portfolio now holds nearly $112 billion in AI-linked companies, about 31%, unde…

Applied Materials, which supplies wafer fabrication equipment to Micron and to Nvidia's manufacturing partner…

Anthropic shares fell 8.5% to close at $97.17 on Friday, down from the $111.50 opening price after the company…

Home Depot says its Magic Apron AI assistant now fields millions of questions per month and is trained on prod…

Berkshire Hathaway has been a popular 'anti-AI trade,' but one firm now says it is too expensive, CNBC reports

Walmart's advertising revenue jumped 25%, driven by traction in AI and TV deals, Yahoo Finance reported
