SoftBank’s relationship with OpenAI has grown from a notable investment into something that now shapes the entire logic of how Masayoshi Son allocates capital – and not always to the benefit of the funds that made SoftBank famous.
A $40 Billion Commitment That Changed the Calculus
When SoftBank committed $40 billion to OpenAI earlier this year, the deal was widely read as Son doubling down on his long-standing conviction that artificial general intelligence would define the next century of wealth creation. What received less attention was what that commitment quietly displaces. SoftBank’s Vision Fund 1 and Vision Fund 2 – the vehicles that brought in outside institutional capital and built positions across hundreds of startups – are effectively being sidelined as Son concentrates firepower directly on OpenAI through SoftBank’s own balance sheet.
The structure of the OpenAI bet matters here. SoftBank is not routing its OpenAI exposure through either Vision Fund. The capital flows directly from SoftBank Group Corp itself, which means the limited partners who invested in Vision Fund 1 – sovereign wealth funds from Saudi Arabia and Abu Dhabi, along with Apple and other institutional backers – do not participate in whatever upside OpenAI eventually generates. That asymmetry is striking given that those same LPs absorbed significant losses during Vision Fund 1’s troubled portfolio years, including the WeWork debacle and the painful markdown cycle that followed the 2021 tech correction.
Vision Fund 2, which SoftBank funded almost entirely with its own money after external LPs largely declined to participate, is in a different position but faces its own quiet marginalization. New deals, new thesis bets, and the headline relationship with OpenAI are all happening above the fund structure entirely. The Vision Funds are increasingly a legacy container holding prior commitments rather than an active vehicle for Son’s current thinking.
This isn’t an accident or an administrative detail. It reflects a deliberate shift in how Son wants to control his most important positions. By keeping OpenAI at the parent company level, he retains full flexibility over timing, governance, and any future monetization without the oversight or distribution obligations that come with managing an institutional fund.
What Gets Crowded Out When One Bet Dominates
The Vision Funds were built on a diversification logic – make large bets across many sectors, accept that some will fail catastrophically, and count on a handful of breakout winners to carry the portfolio. That logic requires active deal-making, a steady pipeline of new positions, and the capacity to allocate across cycles. Right now, the organizational bandwidth and capital attention at SoftBank are visibly oriented around one relationship.
Portfolio companies in the Vision Funds that might once have expected follow-on investment, strategic introductions, or Son’s personal attention during a fundraising cycle are competing with a single AI company that has absorbed more of SoftBank’s identity than any investment since Alibaba. The Alibaba comparison gets made often, and Son himself has invited it, but Alibaba came at a time when SoftBank had no comparable competing priority. OpenAI arrives while dozens of Vision Fund companies are still working through difficult markets and need active support.
The fee dynamics add another layer of friction. Vision Fund LPs pay management fees and carry on a portfolio that is no longer receiving the same quality of attention that justified those terms. A fund manager who is structurally oriented toward deals happening outside the fund raises obvious questions about fiduciary focus. SoftBank has not publicly addressed this tension, and no LP has aired a formal complaint on the record, but the structural logic is difficult to ignore.
There is also a valuation question that sits unresolved. OpenAI’s most recent funding rounds have priced the company at levels that require extraordinary execution on multiple fronts – enterprise adoption, compute cost reduction, regulatory navigation, and competitive defense against Google, Meta, and a growing field of open-source alternatives. SoftBank’s balance sheet is now significantly exposed to that outcome. If OpenAI’s valuation compresses over the next two to three years, the damage lands directly on SoftBank Group, not distributed across a fund structure. That concentration risk is real and is not being offset by the Vision Funds’ diversification, because the Vision Funds are not part of this trade.
Son has shown before that he is willing to accept enormous concentration risk when he believes the thesis is right. His early Alibaba position was held through years of doubt and a dot-com crash that destroyed most of his other bets. The question is whether OpenAI’s path to returns operates on a similar timeline – and whether SoftBank’s balance sheet health and LP relationships can sustain a long waiting period if the AI monetization story takes longer than the current hype cycle suggests.
The Vision Fund’s Shrinking Relevance
SoftBank is still technically running Vision Fund 2 investments and managing the existing Vision Fund 1 portfolio through to exits. But the narrative energy, the capital commitment, and Son’s public statements are all pointing in one direction. The Vision Fund model – which was itself controversial for injecting so much capital into late-stage startups that it distorted valuations across the entire venture ecosystem – may be quietly winding down as a format, replaced by something closer to a holding company that makes a small number of very large direct bets.
That would mark a genuine structural change for SoftBank, not just a strategic pivot. The Vision Funds created a specific set of obligations, a specific investor base, and a specific accountability structure. Walking away from that model while those funds are still active and still carrying LP capital raises a question that Son has not yet answered publicly: at what point does the OpenAI bet become so central to SoftBank’s identity that the Vision Fund era is simply declared over, and what do the people who funded that era get out of it?
