Generative AI Market: Spending, Models and Enterprise Use

This report tracks spending on generative AI (genAI) products and services, and how widely organisations use the technology. It relies on Gartner's public forecast tables, Stanford HAI's AI Index 2025 and McKinsey's 2025 global survey.

Gartner forecast worldwide genAI spending of $644 billion in 2025, an increase of 76.4% from 2024 [1]. In 2024, genAI spending was $365.0 billion, up 336.7% on the year before [1].

Devices are the largest genAI spending category. Gartner's March 2025 forecast put 2025 genAI spending at:

• Devices: $398.3 billion, up 99.5% [1]
• Servers: $180.6 billion, up 33.1% [1]
• Software: $37.2 billion, up 93.9% [1]
• Services: $27.8 billion, up 162.6% [1]

Gartner said about 80% of genAI spending in 2025 goes to hardware, driven by AI capabilities built into servers, smartphones and PCs [1]. It expects AI-enabled devices to make up almost the entire consumer device market by 2028, and noted that consumers are not actively seeking these features [1].

Gartner estimated end-user spending on genAI models at $14.2 billion in 2025, including $1.1 billion on specialized models such as domain-specific language models [4].

In its September 2026 forecast, Gartner puts generative AI models spending at $13.0 billion in 2025, $28.3 billion in 2026 and $51.6 billion in 2027 [2]. It raised its 2026 growth forecast for the segment from 110% to 117%, citing demand for cheaper, domain-specific models [2]. Spending on AI agents and assistants is forecast at $29.2 billion in 2026 and $65.5 billion in 2027 [2].

Stanford's AI Index reports that the share of survey respondents using genAI in at least one business function more than doubled, from 33% in 2023 to 71% in 2024 [3]. Private investment in genAI reached $33.9 billion in 2024, more than 20% of all AI-related private investment [3].

McKinsey's 2025 survey found that 23% of respondents' organisations are scaling an agentic AI system in at least one business function, and a further 39% have begun experimenting with AI agents [5].

Gartner reported that expectations for genAI capabilities were declining in 2025 because of high failure rates in initial proof-of-concept work [1]. It expected CIOs to cut back self-built projects and buy genAI features from existing software providers instead [1]. By September 2026, Gartner placed genAI in the Trough of Disillusionment of its hype cycle [2].

Gartner genAI spending figures are from its March 2025 press release, in millions of US dollars, converted to billions. 2024 figures are reported values and 2025 figures are forecasts. Gartner builds the forecast from sales analysis of over a thousand vendors [1]. The genAI models series is from Gartner's September 2026 AI spending forecast, where 2026 and 2027 are forecasts. Gartner's July 2025 estimate for genAI models ($14.2 billion) differs from the later $13.0 billion figure for 2025 because the forecasts were made at different times.

Adoption figures are survey responses (Stanford AI Index citing McKinsey surveys; McKinsey 2025 survey of 1,993 participants). No FREEE.REPORT calculations are used.