Gartner Forecasts Global AI Spending to Reach $2.7 Trillion by 2026
Gartner predicts that global AI spending will reach $2.7 trillion by 2026, a 49.5% increase from 2025.
The agency lists AI infrastructure as the largest spending category, expected to reach $1.48 trillion in 2026, up from $981.9 billion in 2025; drivers include hyperscale enterprises and service providers expanding data center capacity to support anticipated workloads.
AI services spending is expected to grow from $434 billion in 2025 to $576.5 billion in 2026, with enterprises increasingly leveraging features embedded in existing software through service providers rather than large-scale transformation projects.
AI software spending is expected to reach approximately $461.6 billion in 2026; spending on agents and assistants is expected to rise from $16.5 billion in 2025 to $29.2 billion, with spending on generative models also increasing.
Gartner further predicts that total global AI spending will reach about $3.64 trillion by 2027; analysts note that demand is focused on optimizing infrastructure, including AI-optimized servers, networks, and semiconductors.
Funding is primarily directed towards suppliers of chips, servers, data centers, and embedded software capabilities; hyperscale cloud vendors and hardware suppliers are the main beneficiaries, while traditional enterprises and data center operators face pressure from ongoing spending expansion amid a widening revenue gap that needs to be justified.
Gartner has raised its forecasts multiple times and views infrastructure as the current largest and relatively rigid spending segment.
Source: Public Information
ABAB AI Insight
Gartner continuously tracks global AI spending, with early forecasts raised multiple times, particularly highlighting the infrastructure category; historical technology cycles, such as railroads and internet construction, also experienced a phase of capital concentration on underlying infrastructure.
The capital path reflects hyperscale enterprises and chip manufacturers dominating spending, with resources concentrated on data centers, AI-optimized servers, and power support, strategically locking in future workloads by expanding capacity while embedding AI in software and services as a secondary growth point.
This is similar to the early expansion of cloud computing infrastructure or semiconductor cycles, akin to the railroad era where capital was prioritized for tracks and locomotives; the current phase is characterized by infrastructure-led expansion, with direct enterprise spending not yet fully realized.
Essentially, this represents capital concentration: spending is highly concentrated among a few infrastructure and platform suppliers, with mechanisms in place to lock in demand through capacity pre-positioning and embedded functionalities, directing funds towards scalable hardware and cloud supply rather than dispersed terminal application return verification.
ABAB News · Cognitive Law
Infrastructure spending leads, application returns lag
Speed of forecast upgrades reflects degree of capital concentration
Embedded functionalities expand spending while compressing independent project space