On March 7, Eastern time, according to The Information reports that Mustafa, head of Microsoft's internal AI division When Suleyman took over Microsoft's AI business, he was tasked with a daunting and paradoxical task: to maintain Microsoft's partnership with OpenAI while pushing Microsoft to become autonomous and controllable in the field of AI.

But in this highly competitive market, can Microsoft really get rid of OpenAI's influence on the AI track?

Microsoft and OpenAI: $13 billion "love-hate entanglement"

In Jingtai's view, the partnership between Microsoft and OpenAI is a typical case of "coopetition and cooperation", which is full of strategic synergy and hidden interest games.

At present, Microsoft has invested more than $13 billion in OpenAI, and this deal not only brings Microsoft a portion of the revenue share, but also gives it the right to use OpenAI's technology. However, the cooperation between the two sides has not been smooth sailing, and some recent frictions have exposed deep-seated contradictions.

According to people familiar with the matter, in a video conference last fall, Suleyman, the head of Microsoft's AI division, asked OpenAI to provide detailed technical documentation of its latest AI model, O1, specifically the model's "chain of thought" reasoning process before generating responses. However, OpenAI rejected the request on the grounds of protecting core technical secrets. This response sparked a backlash from Suleyman, who directly raised his voice during the meeting, accusing OpenAI of failing to fulfill the transparency promises in the partnership agreement, and finally ended the call early in anger.

The incident also reveals the delicate relationship between Microsoft and OpenAI: on the one hand, Microsoft relies on OpenAI's cutting-edge AI technology to drive its own business, and on the other hand, it does not want to be locked into a single partner for a long time. Therefore, one of Suleyman's core tasks is to promote the independent development of Microsoft's AI business while maintaining its partnership with OpenAI, so as to reduce its dependence on external technologies.

Self-developed MAI model to challenge OpenAI's hegemony

In order to reduce the dependence on OpenAI, Suleyman, the head of Microsoft's AI division, formed a team led by Karén Simonyan leads an AI research team focused on developing Microsoft's self-developed AI model MAI (Microsoft AI）。

According to internal sources, the performance of the MAI model in a number of AI evaluation standards has been close to or even comparable to the leading models of OpenAI and Anthropic. What's more noteworthy is that Microsoft is developing a model with "chain of thought" inference technology, which aims to improve AI's ability to solve complex problems, which is highly similar to OpenAI's core technology route. This technological breakthrough not only demonstrates Microsoft's R&D strength, but also lays the foundation for its future direct competition with OpenAI.

Currently, Suleyman's team has started working at Microsoft The MAI model was tested in the Copilot product, and the model was gradually replaced by self-developed technology. In addition, Microsoft plans to launch MAI's API as early as later this year, making it available to external developers. This move will give Microsoft direct access to AI API market, head-to-head with competitors such as OpenAI and Anthropic. The launch of the API will not only bring new revenue streams to Microsoft, but will also further strengthen its position in the AI ecosystem.

At the same time, Microsoft is also actively diversifying its technology risks.

Suleyman's team is testing several competitors' AI models, including those from Anthropic, Elon Musk's xAI, and open-source models from DeepSeek and Meta. This diversified technology strategy not only reduces reliance on a single vendor, but also provides Microsoft with more technology choice and flexibility.

MAI model breakthrough and OpenAI's sprint race

Although the Microsoft AI team has made some progress on its own MAI model, the process has not been smooth. The development of the MAI model took nearly a year, during which it experienced many technical bottlenecks and strategic adjustments, and even caused internal senior management dissatisfaction with Suleyman's management style and technical roadmap, resulting in the departure of some core members.

At the same time, OpenAI's technological progress remains impressive. Its continuous release of new models continues to lead in performance and functionality, further widening the technology gap with Microsoft. Nathan, a well-known investor in the field of AI Benaich commented: "From an outside perspective, a year after the establishment of Microsoft's AI division, the actual results are still not clear. They need to make Copilot a serious competitor to ChatGPT, but there seems to be a lack of clarity on the current strategic direction. ”

From the financial data, Microsoft's AI business has exceeded $13 billion in annual revenue, an increase of $3 billion from three months ago, showing strong growth momentum. However, this revenue is mainly dependent on OpenAI-related Azure cloud computing business and Office 365 Enterprise AI Products. In contrast, Suleyman is responsible for consumer-grade AI products such as Bing and Windows AI) is still limited in revenue contribution and has yet to make a significant commercialization breakthrough.

Microsoft AI: Optimistic in the long term, cautious in the short term

In terms of financial performance, Microsoft's AI business has exceeded $13 billion in annual revenue, showing strong growth potential, but its revenue mainly relies on OpenAI-related Azure cloud computing and enterprise-level products, and the commercialization of consumer AI products is relatively slow. At the same time, OpenAI is still leading in the speed of technological progress and market influence, which puts direct competitive pressure on Microsoft.

Nonetheless, Microsoft's long-term strategy is worth watching. Microsoft's goal is not to compete in the market in the short term, but to accumulate technology and build an ecosystem in the next 10 years. This robust strategy, while unlikely to see significant results in the short term, lays the foundation for Microsoft's long-term competitiveness in the AI space.

Optimistic in the long term, cautious in the short term: Microsoft's layout in the AI field has long-term potential, especially the construction of its self-developed technology and ecosystem. However, in the short term, the challenges of technology catch-up and commercialization cannot be ignored, and investors need to remain cautious.

Focus on technological breakthroughs: The progress of the MAI model will be a key indicator. If Microsoft can approach or even surpass OpenAI in performance, it will significantly improve its market competitiveness.

Commercialization progress: Investors should keep an eye on Microsoft's consumer AI products, such as Bing and Windows AI), and MAI The market acceptance of APIs, which will be an important driver for the diversification of Microsoft's AI business revenue.