SoftBank Sends $10B to OpenAI

SoftBank has now sent the first $10 billion of a planned $30 billion follow-on investment into OpenAI through Vision Fund 2. The schedule was set in late February, when SoftBank said the money would go out in three equal tranches on April 1, July 1, and October 1, 2026, and that the deal would buy preferred shares in OpenAI Group PBC at a $730 billion pre-money valuation (group.softbank). This is not a rumor dressed up as strategy. It is a dated funding plan, with the first wire already baked into the paperwork. That matters because SoftBank did not pay from loose cash sitting around. On March 27, it signed a $40 billion bridge facility, explicitly saying the loan was meant primarily to fund the OpenAI investment, with the rest available for general corporate purposes (group.softbank). The lenders were JPMorgan Chase, Goldman Sachs, Mizuho, SMBC, and MUFG, and the loan is unsecured and due in March 2027 (group.softbank). In other words, one of the biggest bets in AI is being financed the old-fashioned way: with borrowed money and a clock ticking. The size of the bet looks even stranger once you line it up against SoftBank’s earlier OpenAI checks. In April 2025, SoftBank said it would invest up to $40 billion in OpenAI, while planning to syndicate $10 billion to co-investors, leaving an effective exposure of up to $30 billion for itself (group.softbank). By February 2026, the company said it had already invested an aggregate $34.6 billion in OpenAI since September 2024 and expected its cumulative investment to reach $64.6 billion after this new round, for an ownership stake of about 13% (group.softbank). SoftBank is no longer treating OpenAI as a portfolio company. It is treating OpenAI as the center of its AI thesis. OpenAI has been inviting exactly that kind of capital because its costs are no longer startup-sized. When the company announced its February 27 fundraising, it said it was raising $110 billion at a $730 billion pre-money valuation, including $30 billion from SoftBank, $30 billion from Nvidia, and $50 billion from Amazon (openai.com). OpenAI framed the need in blunt terms: demand for its products requires compute, distribution, and capital (openai.com). The old software model was to raise money, hire engineers, and ship code. Frontier AI adds a second bill that may be even larger: power-hungry clusters of chips, land, transmission, cooling, and long construction timelines. That is why this funding story keeps turning into an infrastructure story. In January 2025, OpenAI and SoftBank launched Stargate, a new company that said it intended to invest $500 billion over four years to build AI infrastructure in the United States, with $100 billion to be deployed immediately (openai.com; group.softbank). By September 2025, OpenAI, Oracle, and SoftBank said Stargate had grown to nearly 7 gigawatts of planned capacity and more than $400 billion in investment over the next three years, spanning the flagship Abilene site and five new U.S. data center locations (openai.com). Money on this scale is not buying optionality. It is buying electricity, concrete, and time. The corporate structure has shifted to make that easier. SoftBank’s February filing identifies the investment target as OpenAI Group PBC, the public benefit corporation that OpenAI says now houses its for-profit business while remaining controlled by the OpenAI Foundation (group.softbank; openai.com). That change is easy to miss, but it helps explain why these checks keep getting larger. OpenAI is trying to look enough like a mission-driven institution to justify its politics, and enough like a normal company to absorb industrial-scale capital. On April 1, 2026, that hybrid structure took another $10 billion from SoftBank.