Gartner forecasts $4.05 trillion in AI spending for 2027, led by infrastructure
Infrastructure would absorb nearly 60% of the total. Growing workloads are driving the forecast, but higher memory prices are also swelling server bills.
Listen to this story
The audio brief
Story brief
3 key pointsGartner estimates global AI outlays will reach $4.05 trillion in 2027, but its forecast is as much about cost per unit as deployment volume: growing inference and training workloads, cloud-provider build-outs, and memory-driven server price inflation all raise the infrastructure bill. Infrastructure reaches $2.4 trillion, growing faster than the total market, while software and services trail. For investors and operators, server-spending growth should not be treated as equivalent growth in server count or compute capacity.
- 01
AI-optimized server spending is forecast to rise from $527.6 billion in 2026 to $957.2 billion in 2027, about 81%; higher memory costs mean dollars do not map directly to server volumes.
- 02
Spending on AI agents and assistants is projected to more than double, from $30 billion to $67 billion, making it Gartner’s fastest-growing category.
- 03
AI cybersecurity spending is forecast to reach $86 billion, up about 69%; software and services are projected at $649 billion and $746 billion, respectively.
AI infrastructure would consume nearly three dollars of every five spent on artificial intelligence worldwide in 2027 under Gartner’s latest forecast. The firm projects $4.05 trillion in total AI spending, up 47.5%, with growing computing workloads and memory-driven server price increases helping push the bill higher.
The forecast, released at Gartner’s IT Symposium/Xpo in Barcelona, puts infrastructure spending at $2.4 trillion next year, a 53% increase from 2026. That is faster than the projected growth of AI spending overall. Infrastructure remains the largest part of the outlook, even as the firm expects substantial increases in software, services, cybersecurity and AI models.
Gartner identifies two sources of computing demand. Inference—the work of running an already-trained AI model—is scaling faster than anticipated as users’ requirements grow. Training workloads are also becoming more demanding. John-David Lovelock, Gartner’s distinguished vice president analyst and chief of research, attributes the rising infrastructure bill to both trends, rather than to model training alone.
The providers building that capacity are another part of the explanation. Gartner points to continued construction by hyperscalers, or large cloud operators, and neoclouds, the AI-focused cloud providers. Alongside those build-outs, it cites “memflation”: increases in AI server prices driven by memory costs. The forecast therefore combines a growing need for computing infrastructure with a higher price tag for the equipment.
Gartner’s AI server spending forecast
Gartner projects roughly 81% growth in spending on AI-optimized servers between 2026 and 2027.
AI-optimized servers are the largest segment within Gartner’s infrastructure forecast. Their projected spending growth is substantially faster than the 53% increase expected for infrastructure as a whole. But because Gartner explicitly includes rising equipment prices among the drivers, the dollar increase should not be read as an equivalent increase in the number of servers purchased.
Infrastructure dominates the dollars, but agents and assistants lead the forecast’s growth rates. Gartner expects spending on that category to more than double, making it the fastest-growing category in its outlook. The distinction is important: the category with the sharpest percentage increase is not the one attracting the most money.
- AI agents and assistants: Spending is forecast to rise from $30 billion in 2026 to $67 billion in 2027. Despite that rapid increase, the category remains far smaller in dollar terms than infrastructure.
- AI cybersecurity: Gartner projects spending will increase from $51 billion to $86 billion over the same period, approximately 69% growth. That is above the projected growth rate for AI spending overall.
- Generative AI models: Spending is expected to climb 82% to $51 billion in 2027. This is another fast-growing category, but its projected total remains below both cybersecurity and agents and assistants.
Within model spending, Gartner expects domain-specific and specialized models to grow 99% in 2027. Yet foundation generative AI models are forecast to retain 81% of model spending that year. The outlook pairs faster growth for specialized models with continued spending concentration in foundation models; it does not predict that specialization will displace them as the largest spending destination.
Beyond hardware and models, Gartner forecasts AI software spending of $649 billion, up 42%, and AI services spending of $746 billion, up 29.5%. Both totals are substantial, but neither is projected to grow as quickly as infrastructure. In this outlook, the largest spending category is still growing faster than either software or services.
Sources
- itpro.comAI infrastructure spending to top $2 trillion next year
- stocktwits.comAI-Optimized Server Spending To Reach Nearly $957B In 2027, Says Gartner — Global AI Spending To Cross $4T
Reader comments
Newest comments first. Replies stay oldest first.