Corporate AI spending in 2026 is being authorized on the basis of savings that have not materialized. That is the central finding from Bain & Company’s annual corporate-performance survey, reported by Bloomberg, which shows a self-reinforcing cycle: missed targets, renewed optimism, and increased budgets built on projections that the previous year proved wrong.
The numbers are specific. While 86% of surveyed executives expected to hit growth targets last year, 42% missed them, up from 32% the year before. Despite this, companies are forecasting 20% higher revenue growth in 2026 compared to last year, with 91% expressing confidence they will meet those targets. The gap between that confidence and the recent miss rate is not a rounding error. It is the shape of the problem.
Bloomberg’s coverage, citing the Bain release, frames this as an executive-comfort issue. The firm’s language is pointed: the missed targets “should be making executives uncomfortable,” particularly when they are approving AI spend in anticipation of cost savings that AI automation has not yet delivered at scale. Bain identifies a specific binding constraint: 60% of large companies lack the data foundation or the technology infrastructure to scale AI effectively. The problem is not model quality. The problem is that most enterprises cannot yet connect AI outputs to clean, governed, accessible data.
The historical parallel is instructive. Early cloud-migration cost projections in 2015 and 2016 also fell short, and the structural reason was not that cloud technology failed to deliver. The operating model of most large enterprises was not configured to capture cloud’s efficiency gains on the timeline finance teams modeled. IT teams were reorganized after the budget was set. Licensing structures were not renegotiated. Shadow IT persisted. AI deployment in 2026 is running the same pattern: the technology works well enough, but the organizational preconditions, data governance, workflow redesign, and change management were not completed before the savings projections were filed.
One structural note on the source: Bain is a management consultancy. Its business model depends on enterprises purchasing transformation engagements. The same firm publishing the “shortfall” finding sells the services that close the gap. That incentive is worth naming. The survey methodology is still informative, but the framing that enterprises need to buy more consulting to realize AI value should be read with that in mind.
The data-foundation finding is the most actionable signal in the report. Companies that sell AI on top of master data management, data lakes, or governance infrastructure have a structurally easier path to closed deals in 2026 than companies selling model-layer capability. The constraint is upstream of the model. Solving it is where the budget conversation starts.
For B2B founders selling AI tools into enterprise accounts, the expectation-reality gap in this survey is not a warning. It is the sales surface. Budget has been allocated based on ROI projections that did not land. Decision-makers need a vendor who can point to a measurable, near-term savings number, attach it to a specific workflow, and close the delta between what was promised in the 2025 planning cycle and what finance is now asking about. Concrete and measurable beats impressive and vague in every meeting this year.
Reporting by Bloomberg (bloomberg.com), published June 1, 2026, citing Bain & Company’s annual corporate-performance survey.
