Corporate investment in artificial intelligence keeps growing, but turning projects into measurable results remains much harder. Only 22% of organizations have successfully scaled AI across multiple business units or adopted an AI-first approach, according to a Gartner survey of 1,303 functional leaders at companies with at least $50 million in annual revenue.
Key facts about enterprise AI adoption in 20 seconds
- Only 22% of organizations have successfully scaled AI across multiple business units or adopted an AI-first approach.
- 85% of leaders expect to increase their AI spending during 2026.
- In 2025 they devoted an average of 12% of their functional budgets to this technology.
- Productivity is the most sought-after goal, but the most popular use cases don’t always deliver the best returns.
The survey, conducted between January and April 2026, paints a picture considerably less straightforward than the pace of investment might suggest. Companies are putting more money into AI, but a significant share still struggle to know which projects work, what they actually cost, and what economic return they produce.
One of the most striking findings is that about 11% of organizations have no idea at all how much their own function spent on artificial intelligence during 2025.
The problem grows larger because budgets keep rising. 85% of the leaders Gartner surveyed expect to increase spending during 2026, after devoting an average of 12% of their functional budgets to AI-related initiatives last year.
Measuring Returns Makes a Considerable Difference
Gartner’s results show a sharp split between organizations that systematically track their projects and those that still lack clear mechanisms for doing so.
The firm defines high-performing organizations as those that continuously track return on investment (ROI), manage their AI initiatives as a portfolio, and periodically review results to decide where to add resources and which projects to cut or drop.
These organizations reported positive returns on 81% of their AI initiatives.
At the opposite end, low-performing organizations didn’t even know the return on 29% of their projects.
The gap points to a problem that’s gaining importance as AI experimentation gives way to broader rollouts. Approving an assistant, a code-generation system, or an automation tool can be relatively simple. Figuring out afterward how much work it saved, what costs it introduced, and whether it actually improves business outcomes requires upfront metrics and tracking.
Gartner argues that this lack of financial visibility increases risk as spending grows.
Tina Nunno, Distinguished VP Analyst and Gartner Fellow, says that tying investments directly to business outcomes makes it possible to spot underperforming projects earlier and reallocate resources.
The data also shows where much of that money is currently headed.
Productivity Takes Up a Large Share of the Investment
75% of functional leaders named improving productivity as one of the outcomes they’re pursuing through artificial intelligence.
In addition, roughly 30% of the functional budget devoted to AI goes to this goal on average — nearly double the share assigned to the next-largest priority.
The survey data shows, however, that priorities shift depending on the goal being pursued.
| AI Initiative Goal | Leaders Citing It |
|---|---|
| Improve productivity | 75% |
| Cut costs | 55% |
| Increase revenue | 39% |
| Evolve the business | 38% |
| Manage risk | 34% |
| Transform the business | 33% |
| Retain revenue | 33% |
| Improve resilience | 29% |
Productivity clearly leads the rest. That helps explain corporate interest in assistants, task automation, code generation, and tools capable of reducing manual work.
But the study adds an important caveat: the most popular use cases aren’t necessarily the ones that generate the best returns.
Gartner analyzed this gap specifically within IT.
Among the most common projects are cybersecurity threat detection and response — an area where Gartner has flagged AI as reshaping the threat landscape — used by 54%; IT service desk automation, also at 54%; and automated code generation and refactoring, at 44%.
When the question shifts to the use cases with the highest share of leaders reporting positive returns, the ranking changes.
Smart IT asset and cost optimization comes in at 40%, followed by synthetic data generation at 28%, and automated code generation and refactoring at 23%.
These figures don’t mean that 40% is the financial return delivered by asset optimization. They indicate the share Gartner identified for that use case among those associated with positive returns.
The gap between popularity and profitability creates a problem for companies that pick projects mainly because a given AI application is getting a lot of market attention.
A less visible project can deliver an economic result that’s easier to measure if it solves a specific, well-defined problem.
Scaling AI Takes More Than Just Licensing Models
The 22% figure also helps separate two concepts that are often conflated: using artificial intelligence and managing to deploy it broadly across an organization.
A company can have employees using generative assistants, developers working with coding tools, and departments experimenting with agents without having yet managed to integrate these technologies into processes shared across different business units.
The jump between the two situations raises questions of cost, security, data, integration with enterprise applications, access control, training, and results measurement.
It also explains why growing budgets don’t automatically mean adoption has matured.
Gartner’s survey doesn’t claim the remaining 78% has completely failed with artificial intelligence. The 22% figure refers specifically to organizations that have successfully scaled it across multiple units or adopted an AI-first approach. The rest may be at different stages of experimentation, rollout, or expansion.
The sample isn’t representative of all companies either. Gartner surveyed leaders at organizations that recorded at least $50 million in annual revenue in their 2025 fiscal year, so the results are aimed mainly at enterprises of a certain size and can’t be applied directly to small businesses.
Even with these limitations, the study shows a shift in the questions being asked around enterprise AI.
During generative AI’s early years, much of the conversation revolved around which models to use and which processes to introduce them into. As budgets grow, organizations need to answer much more specific questions: how much each initiative costs, what outcome it’s aiming for, how it’s measured, and which projects should stop being funded.
The fact that 85% want to spend more while only 22% have managed to scale the technology widely sums up that gap between investment and maturity well.
Frequently asked questions
What percentage of companies have managed to scale artificial intelligence?
Gartner reports that 22% of surveyed organizations have successfully scaled AI across multiple business units or adopted an AI-first approach.
Will companies increase their AI investment during 2026?
85% of the functional leaders surveyed expect to increase artificial intelligence spending during 2026. In 2025 they devoted an average of 12% of their functional budgets to this technology.
What is companies’ main goal when using AI?
Productivity tops the list of priorities: 75% of leaders identify it as one of the outcomes they’re pursuing. Roughly 30% of functional AI spending goes to this goal on average.
Are the most popular AI projects also the most profitable?
Not necessarily. Gartner finds gaps between the most widely deployed use cases and those where a higher share of leaders report positive returns, particularly within IT functions.
Source: gartner