Worldwide AI-Optimized IaaS Spending to Nearly Double in 2026, Gartner Says

Gartner forecasts the market will reach $42 billion in 2026 and $66 billion in 2027, as inference workloads overtake training as the dominant driver of enterprise AI infrastructure spending.

Worldwide spending on AI-optimized infrastructure as a service is projected to grow 96.4% in 2026 to reach $42.3 billion, according to Gartner. The growth follows an even sharper 180% increase in 2025, when the segment hit $21.5 billion, and is forecast to continue at a slower but still substantial pace into 2027, reaching $66.1 billion at 56.5% growth.

"This growth is driven by continued demand for infrastructure to support large language model training and the rapid operationalization of AI across enterprise applications and workflows," said Hardeep Singh, senior principal research analyst at Gartner.

By comparison, the broader IaaS market — which includes AI-optimized infrastructure as a subset — is growing far more slowly, at 29.3% in 2026 to reach $287.3 billion, according to Gartner's data. AI-optimized infrastructure remains a small fraction of total IaaS spending but is expanding roughly three times faster than the category as a whole.

Inference spending set to overtake training in 2026

Gartner said global spending on AI inference will surpass spending on model training for the first time in 2026, at $23.3 billion versus $19 billion. Inference is projected to account for 55% of total AI-optimized IaaS spending this year, rising to 59% in 2027, according to the firm.

Singh attributed the shift to enterprises moving from model development into production-scale deployment. "As organizations shift from model development to production-scale deployment, fine-tuned and domain-specific models are increasingly integrated into customer-facing and operational systems, requiring continuous, real-time execution rather than periodic training," he said. "This shift is accelerating the cloud consumption patterns and creating sustained demand for AI-optimized infrastructure."

Gartner also pointed to agentic AI as a factor increasing compute demand, noting that multistep, autonomous execution amplifies compute intensity and is helping position inference as the dominant consumption model for enterprise AI infrastructure.

What it means for enterprise IT buyers

The inference-over-training crossover point Gartner is flagging has direct budgeting implications for IT leaders. Training spend tends to be episodic and concentrated around model development cycles, while inference spend scales continuously with production usage — meaning infrastructure costs increasingly move from a project-based line item to an ongoing operational expense tied to how widely AI gets embedded into customer-facing and internal systems.

That shift also raises the stakes around workload placement and vendor selection. As domain-specific and fine-tuned models move into always-on production environments, enterprises will need to weigh latency, data residency and cost-per-inference considerations more heavily than they did during the training-centric phase of AI adoption, when raw compute capacity was often the primary constraint.