Microsoft plans to expand its data center capacity to over 38 gigawatts (GW) by 2032, more than double the current capacity of about 12 GW, with approximately one-third of it to be dedicated to AI chips.
According to Bloomberg, the expansion plan includes both Microsoft’s self-built and leased data centers, excluding the computing power leased from emerging cloud service providers like CoreWeave. Reuters reported on Thursday (September 10) that Microsoft has not commented on this plan.
The 38 GW scale exceeds even the peak electricity demand in the state of New York. With the current capacity at around 12 GW, Microsoft will need to increase its data center capacity by about 26 GW by 2032, where power and land supply will directly impact the construction speed.
Currently, out of Microsoft’s approximately 12 GW data center capacity, around 2 GW is used for AI-specific chips. As per the plan, when the total capacity reaches over 38 GW by 2032, approximately one-third will be allocated to AI chips, equivalent to around 12 to 13 GW.
The significant increase in capacity by Microsoft is driven by the continuous growth in demand for AI and cloud computing. Due to insufficient server capacity, the company had previously been unable to meet the demands of some customers looking to rent AI chip servers, sometimes having to choose between expanding Azure cloud business and retaining servers for its Copilot AI product.
Microsoft’s Chief Financial Officer Amy Hood recently stated that there is very limited new capacity that can be built and put into use in the next 12 months. The company is currently improving the efficiency of existing servers on one hand, while continuing to search for land and power for long-term expansion on the other.
Expanding AI and cloud computing infrastructure requires massive investments. According to Reuters, Microsoft expects its capital expenditure in the first quarter of the 2027 fiscal year to reach $50 billion, with the full-year capital expenditure for the 2026 calendar year estimated at around $175 billion.
In July, the company had anticipated stronger growth in its cloud business compared to market expectations. Previously, investors had been concerned about whether Microsoft’s substantial investment in AI infrastructure would bring sufficient returns, and the robust demand for cloud services has somewhat alleviated those concerns.