Source: Wall Street Journal.
"The Stargate project" has been announced for more than three months, yet SoftBank has not developed a project financing template nor engaged in detailed discussions with Banks, Private Equity investors, and Asset Management companies. Amid the shadow of tariffs, concerns about a Global economic recession are intensifying, along with the emergence of new high-efficiency models, which have impacted the demand for Datacenters. Additionally, price wars are eroding profits, prompting Institutions to reevaluate their investments in AI infrastructure.
The tariff panic in the USA has hindered the ambitions of SoftBank and OpenAI, with the 100 billion USD 'Stargate' project stagnating.
According to a report by Bloomberg on Monday, SoftBank Group's ambitious 100 billion USD U.S. AI infrastructure investment plan - 'Stargate' - is facing serious obstacles, as Trump's tariff threats have stalled financing negotiations, which is a worrying sign for investors hoping for an AI boom.
In January this year, SoftBank founder Masayoshi Son and OpenAI co-founder Sam Altman announced the 'Stargate' plan, promising to deploy 100 billion USD 'immediately' and gradually expand the investment scale to about 500 billion USD over time. However, more than three months have passed without SoftBank establishing a project financing template or beginning detailed discussions with Banks, Private Equity investors, and Asset Management firms.
With the shadow of tariffs looming, high-efficiency models threaten, and price wars eroding profits.
Reports indicate that preliminary talks between SoftBank and dozens of lending institutions and alternative Asset Management firms such as Mizuho, JPMorgan, Apollo Global Management, and Brookfield Asset Management began as early as this year. However, according to insiders, no deals have been reached as financiers reassess datacenter investments in light of increasing economic volatility and cheaper AI services.
Trump's radical trade policy has severely disrupted the Financial Estimates for AI. Reports cite insiders stating that as lending institutions and debt investors avoid high-risk bets, the rising cost of capital, combined with concerns that a global recession may suppress datacenter demand, are dragging out the discussion process.
What is particularly concerning is that everything from Server racks to Cooling Systems and chips may face higher tariffs. According to TD Cowen analysis, the construction cost of datacenters may increase by 5% to 15%, with some operators facing even higher prices. This undoubtedly adds insult to injury for AI infrastructure projects that are already under high cost pressure.
To make matters more complex, a series of more efficient and cheaper AI models are emerging, such as DeepSeek from China, raising questions about how these models will impact the long-term profitability of projects related to OpenAI.
Furthermore, investors are beginning to realize that AI models are not Scarce Resources, and a price war may erode profit margins, thereby diminishing the commercial rationale for supporting large-scale datacenter construction.
SoftBank is determined to push forward, but the prospects are bleak.
Despite facing challenges, SoftBank—responsible for financing the Stargate project—has formed a team of 20 to 30 people within its Vision Fund venture capital division, focused on the Stargate project and accelerating the rollout of large-scale datacenters. This team includes Vikas J. Parekh, the Managing Director for the Americas, who previously led automation and enterprise software investments at the Vision Fund.
Supporters of the Stargate project include Oracle and Mubadala's MGX. SoftBank is also pursuing support from banks and debt investors on a project-by-project basis, including support from Retirement funds and Insurance companies. The plan involves SoftBank providing a small amount of equity—about 10% to 20%—along with significant debt, including mezzanine Bonds and preferred loans.
However, uncertainty is affecting negotiations. What is worrying is that this stagnation may suggest that the entire AI infrastructure boom is cooling down, potentially triggering a chain reaction on the recent AI craze that has driven up technology stocks.
Editor/jayden
