The head of SoftBank Corp. is preparing for a possible slowdown in artificial intelligence data center demand in the United States. Junichi Miyakawa, chief executive of the mobile carrier, said the company may look elsewhere for growth if the market cools, a notable shift in tone from an industry that has spent years racing to build capacity.
SoftBank Group’s mobile unit had been laying the groundwork for neocloud operations in the US, part of a broader push into AI infrastructure. That plan is now being reassessed in light of the Democratic Party’s strong showing in last week’s midterm elections and growing local opposition to data center projects across American communities.
From expansion to caution
For much of the past two years, the dominant story in AI infrastructure has been one of shortage. Cloud providers, startups, and investment funds have scrambled to secure power, land, and chips for new data centers, betting that demand for generative AI would keep rising. SoftBank has been among the most aggressive, backing large-scale projects and using its balance sheet to position itself at the center of the buildout.
Miyakawa’s comments suggest that even the most committed players are now weighing the risk of overbuilding. A correction in US demand would not mean the end of the AI boom, but it could leave some projects underutilized and force companies to spread their investments across more regions.
Why politics and public opinion matter
The Democratic Party’s sweep of the midterms has introduced fresh uncertainty about federal support for energy-intensive tech projects. Data centers consume enormous amounts of electricity, and a changed political landscape could affect permitting, tax incentives, and environmental review timelines.
At the same time, anti-data center sentiment has been rising in parts of the US. Residents and local officials have pushed back against projects that promise jobs but also strain water supplies, power grids, and housing. Those concerns have already delayed or blocked several planned facilities. For a foreign operator like SoftBank, navigating that landscape requires both patience and flexibility.
SoftBank’s broader AI bet
The mobile unit’s caution stands in contrast to SoftBank Group’s wider AI ambitions. The parent company has poured capital into Arm Holdings, the chip design firm whose technology powers most of the world’s smartphones and an increasing share of data centers. It has also partnered with OpenAI and explored a range of AI investments spanning semiconductors, robotics, and enterprise software.
That strategy reflects founder Masayoshi Son’s belief that artificial intelligence will reshape the global economy. Yet even Son has acknowledged that not every bet will pay off quickly. The mobile unit’s more measured stance may reflect a divisional view that infrastructure returns take longer to materialize and depend heavily on local conditions.
What other markets could offer
If SoftBank scales back its US neocloud plans, it could turn to Asia, the Middle East, or parts of Europe where governments are actively courting AI investment. Japan, SoftBank's home market, has limited land and power in major cities but offers political stability and a large industrial base. Several Gulf states have been offering cheap energy and capital to attract data center operators. Southeast Asia is also growing rapidly as cloud demand rises.
The choice will depend on where SoftBank can secure reliable power, favorable regulation, and customers willing to sign long-term contracts. Those factors vary sharply by country and sometimes by city.
The bigger picture for AI infrastructure
SoftBank is not the only company reassessing its AI infrastructure roadmap. Across the industry, executives and investors have started asking whether the current pace of data center construction is sustainable. Some worry that supply is outpacing near-term demand, especially as the largest AI labs build their own specialized facilities and as model training efficiency improves.
A correction would not necessarily be catastrophic. It could simply mean slower growth, lower rents, and more selective project approvals. Companies with strong balance sheets and patient capital, like SoftBank, would likely weather the downturn better than smaller operators.
What comes next
Miyakawa's statement is best read as a warning rather than a retreat. SoftBank Corp. is still positioning itself to benefit from AI growth, but it is no longer assuming that the US will be the only engine. By watching demand signals and political developments closely, the company hopes to avoid committing billions to facilities that may not find enough customers.
For investors, the message is that the AI infrastructure trade is entering a more discriminating phase. The companies that thrive may be those that can move quickly between regions, secure power at reasonable cost, and resist the temptation to build for a boom that may not last forever.
