The global server market delivered its strongest quarter on record in the third quarter of 2025, spotlighting how deeply AI has reshaped enterprise and cloud infrastructure investment. New data from research firm IDC shows worldwide server revenue reached $112.4 billion during the period, a remarkable 61% increase compared with the same quarter last year and another sign that demand for AI compute is accelerating rather than tapering off.
The pace of growth reflects a market that has moved far beyond traditional server refresh cycles. Spending on servers in the first three quarters of 2025 totaled $314.2 billion, nearly double the revenue recorded over the same period in 2024. IDC attributes that expansion primarily to hyperscalers and large cloud service providers building out dense, AI-optimized infrastructure to support model training and inference at scale.
Big Shift Toward GPU Systems
A clear shift in server architecture accompanied the spending surge. Revenue from x86-based systems grew 32.8% year over year to $76.3 billion, while non-x86 servers climbed far more sharply, zooming up 192.7% to $36.2 billion. The outsized growth in non-x86 platforms reflects increasing interest in alternative architectures optimized for AI workloads, where performance per watt and compute density are critical.
Servers equipped with embedded GPUs continued to dominate the market. IDC estimates that these systems now account for more than half of all server revenue, after growing nearly 50% year over year in the quarter. Hyperscalers remain the primary buyers, but IDC also noted early signs of large AI-driven deployments emerging in research and education environments, further broadening demand.
“IDC expects AI adoption to keep growing at an outstanding pace as major vendors continue reporting record orders and showing strong backlogs,” said IDC research director Juan Seminara.
US Leads But Growth Strong Most Everywhere
That demand is playing out unevenly across regions. The United States led global growth, with server revenue rising 79.1% year over year, driven by a 105.5% increase in spending on accelerated systems. Canada followed with nearly 70% growth. China posted a 37.6% increase and accounted for close to one-fifth of global quarterly revenue, while Asia-Pacific (excluding Japan and China), EMEA, and Japan all delivered solid double-digit gains. Latin America lagged, growing just over 4%.
Vendor rankings shifted modestly amid the boom. Dell Technologies held the top position with an 8.3% revenue share, benefiting from strong demand for accelerated servers. Supermicro ranked second with a 4% share despite a year-over-year revenue decline, while Lenovo and IEIT Systems were statistically tied for third place. Hewlett Packard Enterprise rounded out the top five.
Tight Supplies Create Eager Buyers
While server shipments surged, the quarterly numbers reflected that access to GPU is a growing constraint. According to recent industry reports, leading AI developers ranging from Meta and OpenAI are experiencing acute shortages of the graphics processors required to train and run advanced models. Both companies have acknowledged that demand for GPU capacity is exceeding supply, forcing difficult decisions about how compute resources are allocated.
The shortages help explain the intensity of infrastructure spending highlighted in IDC’s figures. Competition for GPU supply is prompting cloud providers and large enterprises to lock in hardware at scale, even as costs rise. For Meta, the pressure has also accelerated investment in custom silicon as a way to reduce reliance on constrained GPU supply.
Taken together, the data points to a data center market increasingly defined by AI. Servers are absorbing a growing share of enterprise capital budgets, while storage and other infrastructure segments grow more cautiously. As long as demand for AI services continues to outpace available compute, the server market’s rapid growth appears set to continue into 2026.
