OpenAI, the creator of ChatGPT, and Microsoft have reached a critical point in their partnership. The dissolution of Microsoft’s exclusive cloud provider role has allowed OpenAI to explore other sources for the computing power needed to support its large language models.
This shift has sparked friction within both companies. A source close to the ongoing discussions revealed that tensions run high, with employees at both OpenAI and Microsoft expressing dissatisfaction. Some feel disconnected from the unfolding changes, while others are deeply involved in the specifics.
Despite these challenges, the collaboration continues. Sources have pointed out that while the conversations are difficult, both parties are still striving to make the partnership work.
OpenAI CEO Sam Altman has pushed Microsoft for additional computational resources. Microsoft has responded by increasing its database operations to meet OpenAI’s needs, though demands for resources appear to be limitless as OpenAI works toward artificial general intelligence (AGI).
This has led OpenAI to turn to alternative providers for computing power, despite its prior agreement with Microsoft. From Microsoft’s perspective, there are concerns about whether these investments will lead to profitability, especially with AI server demands growing rapidly.
OpenAI expects to remain unprofitable until at least 2029, projecting a loss of billions each year. The company’s spending plan for the next few years has ballooned, with forecasts indicating a total outlay of $115 billion—far surpassing initial estimates.
Meanwhile, OpenAI’s flagship product, ChatGPT, seems to be losing momentum. Recent reports show that the app’s user growth has stalled, with 72 million users in recent months. Global app downloads are expected to drop further as competition in the AI market intensifies.
The release of GPT-5, once anticipated to be a major leap toward AGI, failed to meet expectations. This may have dampened user excitement and contributed to the declining interest.
Despite these challenges, Microsoft’s decision to reduce its dependency on OpenAI may benefit both companies in the long run. Over six years, Microsoft has invested heavily in OpenAI, contributing over $13 billion, including a $10 billion funding deal in 2023. This deal promised both companies exclusive access to each other’s technology.
However, in June 2024, a new deal with Oracle allowed for additional cloud support, weakening Microsoft’s position. OpenAI’s need for more computing power led to requests for 4.5 gigawatts from Oracle by December 2024, further complicating the partnership.
As a result, Microsoft’s exclusive cloud provider status ended, although it still holds rights to first refusal and a share of OpenAI’s profits. This development was followed by OpenAI’s $500 billion Stargate project, which marked another step in the company’s strategy to decentralize its operations.
In March 2025, OpenAI expanded further, securing a $22.4 billion deal with Coreweave and establishing new agreements with Google. These moves exemplify OpenAI’s ongoing attempts to lessen its reliance on Microsoft and other single providers.
OpenAI’s rapid growth has attracted major investments from NVIDIA and AMD, with new data centers being built to meet the increasing demand. These projects are raising concerns about the sustainability of such large-scale investments and the potential for a market collapse.
As Microsoft continues to adapt, it aims to become more self-sufficient in the AI field. The company has begun working on its own AI chips and is exploring “off-frontier” AI models, signaling an effort to reduce its reliance on OpenAI’s offerings.
