OpenAI to Shift AI Compute from Microsoft to SoftBank
According to The Information Report on Friday, OpenAI is forecasting a significant shift in the next five years around who it gets most of its computing power from. OpenAI is significantly shifting its AI infrastructure, moving away from Microsoft’s cloud services and toward SoftBank-backed Stargate. By 2030, OpenAI expects 75 percent of its computing power to come from Stargate, marking a shift that carries a lot of opportunity and risk. Though this shift is coming, OpenAI will keep increasing its spending on Microsoft’s data centers in the next few years. However , the company’s operational expenses are poised to increase significantly.
Reports indicate that OpenAI will burn through $20 billion in cash by 2027, marking a significant financial shift from previous years, a massive jump from the $5 billion spent in 2024. By the decade’s end, OpenAI forecasts that running AI models (inference costs) will surpass AI training expenses, marking a significant shift in its computing strategy. This move signals OpenAI’s push for greater independence in cloud infrastructure as it scales its AI models.
Why Is OpenAI Starting to Move Away from Microsoft?
With this move, OpenAI is positioning itself for a world where computing resources are more often distributed. But is this the right move? Moving computing power over from Microsoft (whose Azure powers OpenAI today) to the SoftBank-backed Stargate project is not something that happens overnight; there is a lot of work to be done. OpenAI has leaned heavily on Microsoft’s Azure cloud, but as AI costs have taken off, the company seems to be looking for more control and diversification over its compute resources. There might be several reasons why they decide this.
Microsoft increasing interest in its in-house AI research might lead to strategic conflicts with OpenAI in the future, which might end up resulting in conflicts of interest between the two. To OpenAI, this could be a mandate to secure its long-term independence. In addition, OpenAI’s rising operational outlays — projected to surpass $20 billion by 2027 — necessitate a more fluid funding approach, and SoftBank is famous for its mega tech bets. In addition, OpenAI may want to decrease the reliance on U.S. cloud providers for strategic reasons as well, whether it be aimed at mitigating risks from potential regulatory scrutiny or geopolitical factors.
What It Signals About OpenAI’s Future
In leaning toward SoftBank-backed computing, OpenAI is making a calculated gamble. This could offer more autonomy, tailor-made AI chips, and improved financial flexibility, in other words. However, SoftBank’s track record of putting money into volatile deals (think WeWork) begs the question of whether this is a sustainable partnership in the long term.
And inference costs (i.e., running AI models) are expected to exceed training costs by 2030, so OpenAI needs a long-term sustainable solution. This could blow up in the face of the SoftBank-funded Stargate project if it fails to deliver the same stability and efficiency that Microsoft Azure provides. Ultimately, OpenAI’s pivot away from Microsoft is a high-stakes transition that could determine its trajectory in the A.I. industry. If done right, it could solidify OpenAI’s role as a leading innovator in AI. However, if the transition faces major roadblocks, it could open up new challenges that slow down OpenAI’s momentum in the AI race.
Read More: OpenAI Drops o3 AI Model to Unify AI Strategy with Game-Changing GPT-5
