IDC: Infrastructure spending on AI is expected to grow by 56% to $497 billion in 2026. Arm architecture replaces x86 as the mainstream computing platform.

According to research company IDC, AI infrastructure spending in the first quarter of 2026 reached 89.7 billion US dollars, with Arm architecture-based rack-mounted GPU servers replacing x86 processors to become the mainstream acceleration computing platform. x86 processors are manufactured by Intel Corporation and AMD. Prospects: AI infrastructure spending continues to grow rapidly, with expected breakthrough of 1 trillion US dollars in 2029 IDC has raised its forecast for AI infrastructure spending for the entire year of 2026 to 497 billion US dollars, a year-on-year increase of nearly 56%, mainly due to the increase in capital spending by super-scale cloud service providers and emerging non-GPU AI demand. The research firm currently estimates that the global AI infrastructure market will surpass 1 trillion US dollars in 2029, reaching 1.08 trillion US dollars, and climbing to 1.21 trillion US dollars in 2030, with a five-year compound annual growth rate (CAGR) of approximately 30% starting from 2025. Key points from the first quarter data: AI infrastructure investment enters a new cycle, computing platform undergoes structural transition In the first quarter of 2026, AI infrastructure spending increased by 33% year-on-year; however, after the base was raised, growth gradually returned to normal, with a stable quarter-on-quarter. The report points out that the data for the first quarter confirms that AI infrastructure investment has transitioned from the initial concept validation stage to a continuous years-long capital investment cycle, with industry competition shifting focus from "how much computing power to purchase" to "which platform will emerge victorious." Enterprise technology buyers, cloud service providers, and governments around the world are making long-term decisions about infrastructure locations, budget allocation, and prioritized AI business areas. In the first quarter of 2026, server-related spending was 87.6 billion US dollars, accounting for 97.6% of total AI infrastructure spending for the period. An increasing number of AI demands are beginning to land on infrastructure that does not rely solely on GPU acceleration. Super-scale data center operators are expanding GPU clusters while deploying AI orchestration tools, data pipeline task clusters, and pure CPU inference clusters to control costs. In the first quarter of 2026, the market size of non-x86 (Arm) acceleration servers soared to 53 billion US dollars, higher than the 47.5 billion US dollars in the fourth quarter of 2025 and the 29.8 billion US dollars in the third quarter of 2025. The market size of x86 acceleration servers decreased from 42.7 billion US dollars in the fourth quarter of 2025 and 51.9 billion US dollars in the third quarter of 2025 to 34.6 billion US dollars in the first quarter of 2026. IDC's Global Infrastructure Tracking Director Juan Seminara stated: "The data for the first quarter of 2026 clearly shows that AI infrastructure investment has entered a new stage, with the focus now shifting from the quantity of computing resources purchased to which platform will emerge victorious. We see that sales of x86 acceleration servers have dropped from 52 billion US dollars to 35 billion US dollars in just two quarters, while ARM rack sales have almost doubled. This is not a sudden drop in demand, but rather a result of architectural change. However, a new generation of x86 platforms is about to be released, and the final outcome is still uncertain. At the same time, we see truly AI-driven demands emerging, such as pure CPU inference clusters, artificial intelligence orchestration tools, and storage upgrades with more AI features." The report points out that the shift in server market share began in the fourth quarter of 2025, reflecting that large buyers are integrating around NVIDIA Corporation's NVL72/GB200 level rack-based platform and shifting purchase volumes away from custom x86 rack-based designs. The ultimate market trends will depend on the pace of product iteration by each vendor; industry supply chain bottlenecks persist, and it is still to be observed which architecture route will ultimately prevail. Pent-up storage demand released, resilience of AI investment highlighted The report adds that pent-up storage procurement demand is being released. In the past one to two years, enterprises prioritized budget allocation to GPU and AI servers, repeatedly delaying storage device upgrades, which are now difficult to put off any longer. The report also points out that the long-overdue demand for storage device upgrades, combined with the new just-in-time demand driven by AI business, further highlights the urgency of external storage strategic layout. However, the proportion of AI-dedicated storage in the overall AI infrastructure investment still remains low, at only 2.4%. Regional performance differentiation: Still dominated by the United States, which spent 67.9 billion US dollars in the first quarter, accounting for 75.7% of global total spending, with a year-on-year growth of 30.3%, but the growth rate continues to slow down. China's related spending increased by 9.3% year-on-year to 7.8 billion US dollars, accounting for 8.7% of global spending. In the first quarter of 2026, the Middle East region (mainly Saudi Arabia and the United Arab Emirates) once again achieved the strongest year-on-year spending growth globally, mainly due to government-supported sovereign AI programs and cooperation with leading super-scale data center operators, despite a decline in spending from the record levels in the fourth quarter of 2025. Seminara stated: "Although global economic and political tensions seem to be dragging down other markets, AI investment has shown extreme resilience." In pre-market trading on Wednesday, Arm's stock price fell by nearly 3%, AMD's stock price fell by about 2%, and Intel Corporation fell by 3.4%.