Fortune 500 AI spend up 83%

Enterprise AI spending is still going up fast. But the more important shift is where the money is going. Big companies are moving past the “let everyone try ChatGPT” phase and into the uglier, more expensive work of making AI usable inside real operations. That means governance, data plumbing, identity controls, workflow redesign, and a lot of integration work. The hype cycle is still there — but the budget line is getting more practical. (research.etr.ai) ### What actually changed? The clearest change is that enterprises are spending more even while pulling back on broad, casual access. ETR’s 2026 Tech Trends panel says 89% of organizations expect higher AI spending in the next year, even as license counts fall. KPMG’s Q1 2025 AI Pulse found projected average AI spending over the next (research.etr.ai)ou companies are not backing away from AI. They are getting pickier about where they deploy it. (research.etr.ai) ### Why would spending rise if seats are falling? Because production AI is not a seat-license story anymore. A pilot can run on a few model subscriptions and a demo app. A real deployment needs data pipelines, security reviews, observability, model governance, and systems that connect to ERP, CRM, call centers, claims workflows, or su(research.etr.ai) trimming novelty spend and funding the hard infrastructure underneath. (research.etr.ai) ### Are companies actually using AI at scale? Some are. Many are not — at least not yet. McKinsey says 88% of respondents report regular AI use in at least one business function, up from 78% a year earlier. But it also says most organizations remain in experimentation or pilot mode, and only about one-third report that their companies(research.etr.ai)nterprise-wide value is still uneven. (mckinsey.com) ### So where is the money landing first? In narrow, high-value use cases. KPMG says 73% of respondents use AI to automate workflows across multiple functions, and 53% use it to route critical information between teams. ETR says leaders are favoring mature cloud and data platforms for modular, targeted deployments rather than broad rollout(mckinsey.com) operations, internal workflow routing, software development, and document-heavy back-office work. (kpmg.com) ### Why does this create a services opportunity? Because most Fortune 500 companies are not buying an off-the-shelf “AI transformation” box. They need outside help stitching models into old systems, setting guardrails, cleaning data, and redesigning work around the tools. That is why the spend story increasingly benefits cloud vendors, data-platform comp(kpmg.com)only one layer. The deployment work is the bigger mess — and often the bigger bill. This is an inference from the shift ETR and KPMG describe toward targeted, production-grade execution. (research.etr.ai) ### Which companies are best positioned? The infrastructure names already embedded in enterprise stacks look strongest. ETR’s ML/AI vendor work puts Microsoft Azure Machine Learning and Google Vertex AI near the top on spending momentum, with Databricks, Dataiku, and others also showing strong positioning. ETR’s broader commentary als(research.etr.ai) vendors that already sit near the data, identity, and workflow layers — not just the ones with the flashiest models. (research.etr.ai) ### What is the catch? The catch is that scaling AI exposes every boring problem companies already had. Bad data. Messy permissions. Compliance friction. Weak internal processes. KPMG says 65% of respondents see scaling use cases as a challenge to realizing ROI, up from 33% the prior quarter. ETR says governance, c(research.etr.ai)enterprise complexity. It is surfacing it. (kpmg.com) ### Bottom line? The important number is not just that AI budgets are up. It is that enterprise buyers are acting like AI is now core infrastructure. Fortune 500 companies already represent roughly two-thirds of U.S. GDP, so when that cohort shifts from pilots to production, the spending ripple hits software, cloud, consulting, and labor planning all at once. The easy demo era is ending. The integration era is here. (fortune.com)