Microsoft significantly increases capital expenditure to address rising hardware costs for AI infrastructure, despite ongoing ROI challenges in its AI division.
Microsoft has announced plans to increase its 2026 capital expenditure by $25 billion to cover rising component costs, particularly for AI infrastructure. The company now expects total CapEx to reach $190 billion, with a substantial portion dedicated to addressing memory and storage price increases that have, in some cases, more than tripled since last autumn.
The increased spending reflects Microsoft's commitment to winning the AI arms race despite escalating hardware costs. In the most recent quarter alone, Microsoft invested approximately $32 billion to bring additional compute capacity online, positioning the company to spend another $158 billion between now and the end of 2026. CFO Amy Hood confirmed that next quarter alone, Microsoft plans to spend about $40 billion on hardware and datacenter infrastructure.
"Despite these significant investments, we expect to remain constrained at least through 2026," Hood stated during Microsoft's Q3 earnings call. This acknowledgment of capacity constraints highlights the unprecedented scale of demand for AI computing resources and the challenges in meeting that demand.
The financial metrics reveal some concerning ratios for investors. In the last four quarters, Microsoft has spent roughly $97 billion on infrastructure and equipment to generate $37 billion in annual recurring revenue (ARR) for its AI services. While this represents a 123% increase from the same period last year, the company has yet to achieve a clear return on investment at this scale.
These financial pressures may have influenced Microsoft's recent strategic shift regarding GitHub Copilot. Earlier this week, the company transitioned Copilot from an all-you-can-eat subscription model to a pay-per-token pricing structure. This change likely reflects both the rising costs of AI computation and Microsoft's newfound flexibility in working with various AI models following its updated relationship with OpenAI.
While Microsoft's AI division continues to build infrastructure at a rapid pace, the company's cloud business remains a consistent profit center. In Q3, Microsoft's cloud division generated $54.5 billion in revenue, accounting for more than half of the company's total $82.9 billion in quarterly revenue. This represents a 29% increase compared to the same period last year, contributing to a 23% year-over-year jump in overall profits to $31.8 billion.
Conversely, Microsoft's personal computing division, which includes PCs, gaming, and Bing, experienced a 1% revenue decline to $13.2 billion. Windows sales fell 2% year-over-year, while Xbox content and services revenues dropped 5%. Improved performance from Bing search helped mitigate some of these losses. Looking ahead, Hood projected Windows OEM revenue to fall in the mid-teens for the next quarter.
Despite these challenges, Microsoft maintains a positive overall outlook. The company expects Q4 revenues to rise 13-15% year-over-year, reaching between $86.7 billion and $87.8 billion.
Microsoft's aggressive spending on AI infrastructure underscores the company's determination to maintain its competitive position in the rapidly evolving AI landscape. The scale of these investments, while presenting near-term financial challenges, positions Microsoft to capitalize on what appears to be an accelerating demand for AI-powered services and infrastructure.
Comments
Please log in or register to join the discussion