Most companies spending on AI still cannot point to it in their earnings, and Forrester expects that to hold through 2027.
In its State of AI Survey 2026, the research firm found that 22% of AI decision makers reported a positive effect on earnings over the past year. More than half of the CIOs surveyed expect to miss their AI ROI (return on investment) targets next year. Forrester still projects AI will make up a fifth of all business technology spend in 2027, as companies move from experiments to more deliberate deployments.
The patience gap is the interesting part. According to the survey, 45% of CIOs want to see a return within one year before they call a rollout a success. Set that beside a 22% rate of earnings impact and the pressure on budget holders is easy to see. Forrester predicts companies will get better at tying AI spending to financial outcomes, but that is a forecast, not a result.
Money is tight too. Most tech leaders expect bigger budgets, yet only 53% expect the increase to beat inflation. Pricing is moving the wrong way for them: many AI providers are shifting from seat-based subscriptions to consumption-based models, where customers pay per token (a small chunk of text the model processes). Metered billing worked this way in cloud computing as well, and it moves the risk of an unpredictable bill from the vendor to the buyer.
Forrester's answer is open-weight models, whose trained parameters are published for anyone to download and run. It expects Gemma, Llama, and Qwen to gain ground because they are "good enough for the enterprise workloads that generate the most token volume," such as document processing, customer support conversations, internal knowledge search, and routine substeps in agentic workflows.
Forrester also says a quantum computing breakthrough could arrive as soon as next year. Among businesses familiar with quantum, 87% have no resources allocated to it. If the breakthrough comes, it will land on budgets that are already stretched, with no money set aside for it.