Global expenditures on artificial intelligence are entering an unprecedented phase of expansion. According to the updated third-quarter 2026 forecast released by market research firm Gartner, total worldwide spending on AI is projected to reach between 2.67 and 2.70 trillion US dollars this year. This represents an increase of 49.5 percent compared to 2025, when total spending stood at 1.79 trillion US dollars. Looking further ahead, Gartner expects global AI investments to accelerate to 3.64 trillion US dollars in 2027.
The overwhelming majority of this capital is being channeled into the physical foundation of the technology rather than consumer-facing software. Compute and semiconductor infrastructure alone accounts for nearly 1.48 trillion US dollars, representing approximately 56 percent of the entire global AI outlay. Hyperscalers and corporate data center operators continue to pour immense resources into advanced graphics processing units, server racks, and high-performance network hardware. This allocation highlights how capital-intensive supplying computing capacity at scale remains.
A striking metric revealed in the forecast is the pronounced structural imbalance between foundational models and raw hardware. For every single US dollar allocated directly to generative AI models, which are expected to generate 28.3 billion US dollars in spending in 2026, enterprises and hyperscalers invest more than 52 US dollars in baseline infrastructure. While public interest centers on model architectures, the financial reality remains anchored to silicon and physical data facilities. The algorithmic tier accounts for only a minor fraction of the current economic commitment.
Adjacent enterprise markets are expanding alongside this hardware boom. Spending on specialized AI services is projected to reach 576.5 billion US dollars in 2026, while software dedicated to AI environments will account for 461.6 billion US dollars. Corporate leaders increasingly depend on external consultancies, systems integrators, and software frameworks to embed generative capabilities into operational pipelines. These expenditures prove that integrating models and adjusting legacy architecture carries a substantially higher price tag than basic model licensing.
Gartner's data paints a clear picture of an AI expansion dominated by physical capacity expansion. While corporate narratives celebrate front-end efficiency and creative assistants, the real commercial battle is fought over electricity grids, chip allocations, and cooling infrastructure. Over the coming fiscal cycles, enterprises will face intensifying scrutiny regarding whether operational productivity gains can materialize quickly enough to validate this massive wave of upfront infrastructure spending.