Microsoft is building its own AI. Does the stock have potential after the worst quarter since 2008?

Microsoft plans to develop its own large language models underpinning artificial intelligence by the end of 2027. The company aims to build its advanced AI solutions as an alternative to what is offered by OpenAI, in which Microsoft has invested more than $13 billion, and Anthropic. Investors are still pessimistic about the company's prospects: the stock has fallen in price by 23% in 2026, last quarter was the worst for the securities in 18 years. But some analysts are still giving the securities a "buy" recommendation.

Microsoft's goal is to reach the most advanced level in artificial intelligence models that can respond to or generate text, images and audio no later than the end of 2027, Microsoft AI chief Mustafa Suleiman said in an interview with Bloomberg.

Microsoft is ramping up its computing power in order to create generalist AI models. To do this, the company began using a cluster of Nvidia GB200 chips in October 2025, and over the next 12 to 18 months will expand the infrastructure to the level needed to create advanced AI models, Suleiman said.

Until recently, the development of its own universal AI models was constrained by the terms of its partnership with OpenAI: Microsoft could not create competing solutions in exchange for access to ChatGPT technologies, Bloomberg noted. This restriction was lifted after the agreement was renegotiated in October 2025, the agency added.

Shares of Microsoft at the auction on April 2 rose by 1.1% - up to $ 373.46. However, compared to the beginning of 2026, the securities are now 23% cheaper. For comparison: the "technological" index Nasdaq Composite has fallen by 6% since the beginning of the year. The first quarter of the year was the worst for Microsoft shares since 2008: investors are concerned about its prospects in the AI market, CNBC wrote.

Meanwhile, 92% of analysts tracking Microsoft stock are optimistic about the stock's prospects, MarketWatch wrote, citing data from FactSet. The stock got a new buy recommendation this week, with Benchmark's Yi Fu Lee initiating their analyst coverage with a Buy rating and $450 target price, MarketWatch reported. Benchmark's target is 20% above the current price. The selloff provides investors with an "attractive buying opportunity," Yi Fu Lee asserted.

One of Microsoft's key advantages is the huge amount of proprietary corporate and user data generated by its ecosystem of products, from Microsoft 365 and Teams to LinkedIn, according to Benchmark, which he calls a "treasure trove" of data that could become the basis for the development of AI. The analyst also emphasized that Microsoft is present at all levels of the technology vertical, from Windows and GitHub to cybersecurity solutions, which will allow it to make money on AI implementation in several market segments at once.

Benchmark also views the partnerships with OpenAI and Anthropic positively, calling them mutually beneficial: model developers get access to the company's infrastructure, and Microsoft gets access to advanced AI technologies.

Microsoft is now releasing narrowly focused AI tools. For example, on April 2, Microsoft unveiled a new speech recognition model: the company claims it outperforms competitors' solutions in terms of accuracy in a number of languages and will be integrated into the Teams service and the Copilot AI agent. However, such developments are trained on smaller amounts of data than OpenAI's GPT-4 or Anthropic's Claude 3 Opus, Bloomberg notes.

In March, Microsoft narrowed the focus of Suleiman's job: he joined 2024 to lead the introduction of AI into the company's consumer products, but now focuses on developing AI models. Development of the Copilot assistant for enterprise and individual users was handed over to former Snap top executive Jacob Andreu. This was an acknowledgement of Microsoft's failure to create a ChatGPT competitor for a wider audience, Bloomberg believes.

This article was AI-translated and verified by a human editor