Technology stalwart Microsoft (MSFT 0.28%) released impressive second-quarter fiscal 2025 financial results on Wednesday, Jan. 29, with revenue and earnings surpassing analyst consensus expectations. Revenue was up 12% year over year to $69.6 billion, while adjusted earnings per share (EPS) rose 10% year over year to $3.23.
Despite the robust performance, Microsoft's shares are down by nearly 9% since the earnings release (at the time of this writing). Investors are disappointed with weaker-than-expected revenue guidance for the third quarter of fiscal 2025. The company is also seeing a slowdown in its Azure cloud services business, triggered by capacity constraints.
And it has been hurt by the marketwide turbulence caused by the announcement of Chinese start-up DeepSeek, an artificial intelligence (AI) model with capabilities claimed to be comparable to OpenAI's GPT-4 at almost 95% lower costs. Doubts are being cast on DeepSeek's claims, but investors have also become concerned about the potential sustainability and lifetime value of Microsoft's AI investments.
Although these worries should not be ignored, the extent of the negative investor reaction seems disproportionate to the actual headwinds. Considering Microsoft's fundamentals, it has become an attractive cloud computing stock on the current dip.
Here's why Microsoft can be a compelling pick for astute investors in 2025.
Overblown DeepSeek fears
Many industry analysts believe that market fears -- triggered by DeepSeek's claim to be able to AI train models at dramatically lower costs with only 2,048 H800 GPUs versus nearly 25,000 H100 GPUs for GPT-4 -- seem overblown. CEO Alexander Wang of Scale AI has hinted that DeepSeek is using as many as 50,000 H100 chips, a fact that the Chinese company seems to have ignored disclosing due to U.S. export controls.
And Martin Vechev, the director of Bulgaria's Institute for Computer Science, Artificial Intelligence, and Technology, has called DeepSeek's cost claims misleading since the reduced number of H800 chips was used for only "one training." However, to develop a robust model, it has to be trained multiple times.
DeepSeek also seems to have excluded costs associated with data collection and other processes. As training 2,048 H800 GPUs one time involves more than $5 million to $6 million; meanwhile, the chips themselves cost between $50 million to $100 million.
Lastly, besides hardware expenses, DeepSeek has also not disclosed salary expenses -- especially important since it is known to have hired talent from prestigious universities in mainland China.
Hence, DeepSeek may not be as big a challenge to established technology players as previously thought.
Strong cloud computing and AI business
Microsoft's cloud computing business continues to be a major growth catalyst. Cloud revenue was $40.9 billion in the second quarter, up 21% year over year. This growth was mainly driven by a 31% year-over-year jump in revenue from the Azure cloud computing platform and other cloud services.
The company's AI business also demonstrated strong momentum and recorded an annual revenue run rate of $13 billion at the end of the second quarter, up 175% on a year-over-year basis.
Azure is the second-leading player in the global market for cloud infrastructure services, with a 24% share in the fourth quarter of 2024 (ending Dec. 31, 2024), up almost 2 percentage points year over year.
Increasing demand for generative AI and other technologies has been crucial in driving up cloud spending. Businesses are increasingly focusing on companywide AI deployments, not just initial proof of concept, to maximize their return on investment in AI initiatives. Microsoft has almost doubled its data center capacity in the last three years to benefit from the dramatic expansion in short-term and long-term demand for AI services.
The company has also recently amended its OpenAI partnership, giving it the choice to move from being an exclusive cloud provider to having the right of first refusal. And beyond OpenAI, Microsoft is also rumored to be investing in other prominent AI companies such as Anthropic, which gives it access to a large portfolio of AI technologies.
The company is also focused on developing in-house capabilities such as chips and models. All these initiatives have made it easier to adapt its AI strategy to changing market conditions.
Valuation
Microsoft is trading at 31.85 times forward earnings, which may seem rich initially, but several reasons support this valuation.
Analysts expect the company's revenue and EPS to grow year over year by 13% and 11.5%, respectively, in fiscal 2025. Although modest, these numbers hint at robust growth and margin trends for a company of Microsoft's scale and broad geographic presence.
The business also enjoys exceptional long-term revenue visibility. Its commercial remaining performance obligations (RPO), a metric of future revenue from existing contracts, were $298 billion at the end of the second quarter, up 34% year over year. Commercial bookings also rose by 67% year over year at the end of the second quarter. The high revenue predictability is considered a solid positive for any company.
Microsoft is also committed to investing nearly $80 billion in AI-enabled data centers, with almost half of the investment scheduled for fiscal 2025. The company had $22.6 billion in capital expenditures in the second quarter, of which over half was attributed to long-lived assets that management plans to monetize over the next 15 years. The sustained long-term growth plan further justifies the company's valuation.
Against this backdrop, Microsoft seems like a smart buy now.