Gartner forecasts that global spending on artificial intelligence (AI)-optimised infrastructure as a service (IaaS) will reach 37.5 billion dollars in 2026, up from 18.3 billion dollars in 2025. The research firm expects this market to expand by 146% by the end of 2025 as demand increases for cloud infrastructure designed for AI workloads.
According to Gartner, AI-optimised IaaS refers to cloud infrastructure that uses high-performance computing components such as graphics processing units (GPUs), application-specific integrated circuits (ASICs), and other AI accelerators to support large-scale AI processing. The company stated that growth in this segment is expected to exceed that of traditional IaaS over the next five years.
Hardeep Singh, Principal Analyst at Gartner, said that as organisations expand their use of AI and generative AI, they will require infrastructure capable of handling high-speed processing and data transfer, including GPUs, tensor processing units (TPUs), and optimised storage systems. Singh noted that traditional central processing unit (CPU)-based IaaS offerings may face challenges in addressing these requirements.
Gartner’s analysis indicates that as AI use becomes more widespread, inferencing workloads will overtake training workloads as the main driver of AI-optimised IaaS demand. The firm projects that spending on inference-focused applications will rise to 20.6 billion dollars in 2026, compared with 9.2 billion dollars in 2025. By 2026, 55% of AI-optimised IaaS spending is expected to support inferencing workloads, increasing to more than 65% by 2029.
Singh explained that inference involves continuous processing that supports real-time applications such as chatbots, recommendation systems, and fraud detection tools, while training requires more intensive compute cycles during model development and updates.
