Omdia anticipates AI infrastructure spending will peak as a share of total data center investment in 2026, before gradually easing through the end of the decade
In sum – what to know:
Market to hit $286B by 2030 – Omdia projects strong long-term demand, though the pace of expansion is slowing after a surge between 2022–2024.
AI spending peaks in 2026 – AI will dominate incremental data center investment mid-decade before gradually tapering off toward 2030.
Competition expands beyond Nvidia – Custom ASICs, merchant ASSPs, and AMD GPUs are gaining traction as efficiency and model design shift computing demand.
The AI data center chip market will reach $286 billion by 2030, according to Omdia’s latest AI Processors for Cloud and Data Center Forecast projects
However, Omidia highlighted in its report that the growth in the AI chip market is starting to slow.
The surge in AI infrastructure demand led to a major upward revision: $123 billion in GPUs and AI accelerators shipped in 2024, rising to $207 billion in 2025 and $286 billion by 2030, according to Omdia.
While shipments grew more than 250% between 2022 and 2024, year-on-year growth is expected to slow to about 67% from 2024 to 2025. Omdia anticipates AI infrastructure spending will peak as a share of total data center investment in 2026, before gradually easing through the end of the decade.
“The key drivers are the adoption of AI applications, broader access to fine-tuning, and reasoning models that produce vast amounts of unused tokens,” said Alexander Harrowell, principal analyst for advanced computing at Omdia. “At the same time, smaller specialized models and efficiency gains in training and inference are curbing demand for compute.”
The report showed that Nvidia continues to dominate the market, but 2025 is seeing increased traction for GPU alternatives. These include custom ASICs like Google’s TPUs, merchant ASSPs such as Huawei’s Ascend, Groq, and Cerebras, and AMD’s Instinct GPUs, strengthened by major software investments in 2024.
