We Tested the Year’s Ten Biggest AI Claims Against Our Own Data

Today we published The Executive State of AI 2026, the annual flagship of the Enterprise AI Buying and Budget Index. It does something we have wanted to do all year: take the ten most-quoted claims in AI research, the numbers that show up in every board deck, and put our own first-party data next to each one, with a verdict. Corroborated, contradicted, or complicated.

The reports we tested are the ones executives actually cite: BCG's finding that 72 percent of CEOs call themselves the main AI decision-maker. MIT's 95 percent of pilots with no measurable return. Gartner's forecast that over 40 percent of agentic AI projects will be canceled by the end of 2027. McKinsey on scaling, PwC on CEO returns, a16z on budgets going core, S&P Global on abandonment, Menlo Ventures on enterprise spend, KPMG on agents, OECD on where the venture money went.

Three verdicts stand out. BCG's 72 percent is real, and it is a self-report: our seat cuts show CEO respondents naming the CEO at 84 percent while technology respondents say 24 percent. The closer the respondent sits to the work, the smaller the CEO's claimed role gets. MIT's 95 percent reads as a deployment failure, but our rooms say deployment succeeded and measurement did not: 71 percent of the largest finance room already runs an AI tool, and proving ROI is the top blocker at 51 percent. And Gartner's cancellation forecast finds no retreat in our data, 81 percent of marketing and growth respondents are past exploring agents, but the conditions it depends on are real: only 36 percent of security teams have a dedicated AI security budget line.

Then there is the claim nobody made, because no external survey asks sellers and signers the same question. The CEO is the most-named signer of AI purchases. Founders selling AI pitch the business unit and, increasingly, the CFO. And the seat that has to secure what gets bought was named by zero of 148 founders in our July cohorts. We call it the seat gap, and first-party operator data is the only place it shows up.

The report also carries the series constructs in one citable place: the Self-Attribution Effect, the Ownership Vacuum ("no single owner yet" doubled to 14 percent inside July, and zero CEO respondents report it), the 27-point Optimism Gap between the seat that signs and the seat that proves, and the Executive AI Leverage Ladder. Plus a 25-question FAQ mapped to the questions the major AI reports get asked, answered with our data where we have it and honest links where we do not.

This is an early edition, built on one quarter of Index data, and it says so in print. Fuller editions follow as the monthly cycles accumulate. Read it at openfutureforum.com/research/executive-state-of-ai, cite any figure from the statistics page, and if you would rather be in the next dataset than read about it, the calendar is here.

Last updated: August 1, 2026

Murray Newlands

Founder, Open Future Forum

Murray Newlands has been building executive communities in Silicon Valley since 2019. Open Future Forum hosts private dinners and events for C-suite leaders and board directors navigating the AI era, grounded in a give-first philosophy.

About Murray LinkedIn Twitter / X Substack