Gartner forecast on September 16 that worldwide AI spending will total $2.7 trillion in 2026, a 49.5% increase over 2025. Infrastructure, spanning AI-optimized cloud services, servers, network fabric, and processing semiconductors, accounts for more than half of that total, with AI services at $576.5 billion and AI software at $461.6 billion behind it. Gartner distinguished VP analyst John-David Lovelock described the scale directly: “The buildout of AI data center capacity is the largest infrastructure project humanity has ever undertaken.”
What the breakdown shows
Gartner’s figures put a specific ratio on a pattern enterprise buyers have been observing anecdotally: for every dollar spent on generative AI models themselves, more than $52 is being spent on the infrastructure underneath them. Generative AI models are forecast to grow 117% in 2026, the fastest rate in the forecast, but off a small enough base that infrastructure still dwarfs it in absolute dollars. Gartner has raised its 2026 AI spending forecast by roughly $143 billion since January, and about 83% of that increase came from a higher infrastructure forecast specifically, not from services or software.
Why it matters to enterprise AI buyers
The $52-to-$1 infrastructure-to-model ratio is a useful gut-check for any organization budgeting its own AI program: if internal spending skews heavily toward model licensing or API costs relative to the compute, storage, and networking underneath, that allocation runs counter to where the broader market is actually putting its money, and may signal underinvestment in the infrastructure layer that makes model spend usable at production scale. The consistent upward revision pattern, this is Gartner’s third raised AI-specific forecast tracked this year, is also worth building into budget planning as a bias rather than treating any single number as final.
Source: Techstrong.ai / AIwire
