The Uncomfortable Truth Behind the Enterprise AI Spending vs Adoption Report
Budgets keep climbing while results lag behind.
The enterprise AI spending vs adoption report numbers show spending running ahead of adoption. Peter White at Automation Anywhere reported that 86% of AI investment funds copilots and assistants, and Stefan Weitz, CEO of HumanX, noted that global AI spending reached $2.5 trillion. White observed that only a small fraction of enterprises see significant revenue impact.
Sven Peters at Atlassian recalled that the company hit its goal of doubling AI usage within six months. The top-line number hid a split: engineering drove it, while legal and marketing ignored the tools. Peters admitted Atlassian set the wrong incentives by measuring usage at the company level, which masked who was actually working differently.
Why do vanity AI metrics mislead?
Peters reported that only 6% of executives see a return on their AI investments. Atlassian first tracked company-wide super users who engaged with AI 40 times a week. Once the team saw that this number hid dead zones in legal and marketing, it switched to tracking super users per function.
To drive adoption function by function, Atlassian built hero use cases: call prep agents for sales, intelligent document processing for finance, and an HR agent that builds onboarding plans for new employees. Peters also described an agent that analyzes themes across customer interviews and outputs a Confluence whiteboard in place of a text document, which forces people and agents to work on it together. He treats numbers like 27,000 freed hours as signals only.
What does the enterprise AI spending vs adoption report say about copilots?
White said 86% of AI investment today goes into copilots or assistants, and a very small fraction of the 88% of companies using enterprise AI report a significant revenue impact. Tool sprawl makes it worse. A large enterprise might run over 900 applications, White observed, and 40% of them now offer agents. He added that 98% of companies report employees using AI tools that IT has not sanctioned.
One panelist running a small media organization admitted they buy a new subscription every day and realize 15 days later they made a mistake, then revert.
Weitz reported that 95% of AI pilots fail to reach production and that 84% of companies have not designed a single workflow. Companies are spending $2.5 trillion to bolt AI onto org charts and processes designed for a very different world.
Why does bolting AI onto broken processes fail?
Weitz cited a friend's study of 20 companies deploying AI agents. Fourteen failed because they automated a fiction of how they thought the company worked, instead of the real processes.
Skipping the dull infrastructure work to chase new models burns cash. One panelist recalled an early customer that spent $120 million on coding agents. Stale knowledge in Confluence, Notion, and other systems sent the agents down wrong paths, and the customer wasted over half of its token spend.
Phil Wiser, CTO at Paramount, argued that current AI solutions fail because they rely on low-dimension information, such as ERP financial system data, to understand how a company operates. When a company rolls out a large-scale financial system, Wiser said, 70% of the time it fails.
How do leading enterprises prove AI value?
Francis deSouza, COO of Google Cloud, recommended picking five to seven use cases a year and driving adoption from the top down. Companies that let 1,000 flowers bloom often end up with 1,000 dead flowers, he warned. In Google's treasury department, deSouza said, Gemini optimizes daily cash balances and has generated tens of millions of dollars of benefit.
At Mercedes-Benz, Katrin Lehmann recalled that the CEO built his own agent to understand the technology and encourage everyone else. Every Friday morning, the whole organization drops everything and learns.
Where does human judgment still matter?
Alison Moore at Chief reported that 87% of women leaders have witnessed negative outcomes when AI is prioritized without equal investment in people. She recalled a chief product officer whose team uses AI to analyze customer feedback, while the decision about which product features to cut stays 100% human, because the stakes are high and the trade-offs are too nuanced to automate.
Moore also recalled a CRO who restructured her sales team so AI handles account research and proposals. Every new sales rep still spends the first six months building customer relationships the old-fashioned way.
Peters argued that people should use AI to challenge their thinking and keep their own thinking in the loop. He recalled a colleague who had AI write an event strategy just to get a first reaction, which cost time and trust.
The enterprise AI spending vs adoption report numbers point to a clear order of work: fix the process, connect the data, pick a few use cases, and measure by function.
To hear more from the people putting AI into production, see the HumanX 2027 program preview.
Frequently asked questions
What are the main reasons enterprise AI spending is growing faster than adoption rates?
Stefan Weitz at HumanX reported that companies are spending $2.5 trillion to bolt AI onto outdated processes, and Peter White at Automation Anywhere noted that 86% of AI investment goes into copilots and assistants. Weitz added that 95% of AI pilots fail to reach production because organizations end up automating a fiction of how they think they work instead of their actual processes.
How can CTOs and CISOs demonstrate AI ROI to the board amid slow adoption?
Francis deSouza at Google Cloud recommended picking five to seven use cases a year and driving them from the top down, warning that letting a thousand flowers bloom often leaves a thousand dead flowers. He pointed to Google's treasury team, which uses Gemini to optimize daily cash balances and has generated tens of millions of dollars of benefit.
What frameworks do leading enterprises use to measure AI project success?
Sven Peters at Atlassian argued that companies should measure super users per function and skip company-wide usage metrics. Atlassian found that its company-wide number hid the fact that legal and marketing had abandoned the tools, so it moved to tracking engagement by department.
How do enterprises avoid AI tool sprawl and ensure effective adoption?
Katrin Lehmann at Mercedes-Benz said the company reserves every Friday morning for the whole organization to drop everything and learn, with experts running hands-on training so employees can build their own agents. She added that leadership has to understand the technology, and that even the chief executive built his own agent.
What evidence should be required from AI vendors to substantiate their claims?
David Swan at Retool argued that vendor tools should connect to real systems of record and pass unit testing. He reported that only 1% of solutions currently meet that production bar, and that AI deployments still need traditional software development methods. He also said solutions should fit the user's problem and avoid adding to the backlog.