Gartner sees $2.52T global AI spend by 2026, led by infra buildout

Gartner’s $2.52 trillion AI spending forecast for 2026 is less about a precise number and more about the order of magnitude: we are moving into a world where AI is treated as a foundational capital expenditure category, not a line item in IT budgets. The emphasis on AI-optimized servers and a $401 billion bump in infrastructure outlays shows that most of this spend is still going into the plumbing—GPUs, networking, storage and data center upgrades—required to train and serve ever larger models.([all-about-security.de](https://www.all-about-security.de/gartner-prognostiziert-weltweite-ki-ausgaben-von-25-billionen-us-dollar-fuer-2026/))

From an AGI perspective, trillions of dollars of cumulative spend over the next few years make it increasingly likely that someone will be able to afford the experiments required to push beyond current scaling laws plateaus, whether via bigger models, better data, new architectures or agentic systems. The catch is the “phase of disillusionment” Gartner flags: enterprises are becoming more demanding about demonstrated ROI, not just proofs of concept. That tension—between investor and boardroom skepticism and a macro trend of massive capex—is exactly the environment in which the leading AI labs and cloud providers will be pushed to deliver more capable, more general systems faster, but with clearer economic narratives.