Enterprise AI Adoption: Pilots, Production and ROI — 2026-09-28
This week's headlines centre on widening gaps between AI spending and measurable returns: a new Dataiku/Harris Poll survey finds 81% of global CIOs say they have lost oversight of AI agents, while Gartner now projects AI spend will jump 49.5% in 2026. McKinsey warns agentic AI will make AI costs more variable and expensive, and CFOs are pushing back harder as 2027 budget planning begins. In Japan, a PRTimes survey of retail and food-service firms highlights the tension between labour shortages and AI adoption realities.
Enterprise AI Adoption: Pilots, Production and ROI — 2026-09-28
Top developments
CIOs admit loss of oversight as AI agents multiply
A new Dataiku/Harris Poll survey of 685 global CIOs, published this week, finds 81% say they have lost oversight of AI agents, 84% say employees build agents faster than IT can govern them, and 72% cannot consistently measure whether their agents deliver business outcomes. For pilot-to-production conversion, this is the core problem: untracked, ungoverned agents make ROI unprovable at exactly the moment CFOs demand evidence.
Gartner: AI spend to jump 49.5% in 2026 — "every dollar an AI dollar" by 2030
Reporting published this week says Gartner projects AI spending growth of 49.5% in 2026, with one Gartner analyst stating that by 2030 "every dollar spent will be an AI dollar." Separately, CIO.com reports today that AI spending is squeezing out IT modernization budgets, delaying the foundational data-platform work that underpins measurable AI ROI — a direct risk to pilot-to-production conversion rates in 2027 planning.
McKinsey warns agentic AI makes AI costs less predictable
Business Insider reported this week that McKinsey is warning companies AI spending may rise further and become less predictable with the adoption of agents — agent workloads create variable, usage-based cost structures rather than fixed licences. This shifts the budget-line conversation: enterprises budgeting for 2027 need FinOps-style AI cost tracking, and unpredictable costs are becoming a stated reason CFOs delay production rollouts.
The 2027 budget gate: CFOs demand proof before funding AI
As 2027 planning begins, coverage this week from Innover Digital describes a new "AI approval gate" — trust in AI spending has fallen even though spending itself hasn't slowed, and CFOs want CIOs to prove returns before funding further scale-up. Per-employee enterprise AI spending is now estimated at $2,068 in 2026, up more than 50% from $1,358 in 2025, raising the bar for what measured returns must cover.
TechTarget: models got better, ROI didn't
TechTarget argued this week that despite improved AI models, enterprise ROI remains stagnant, and that organizations must redesign workflows — not just upgrade models — to unlock value. The argument reframes the abandonment debate: stalled pilots are often a process-design failure rather than a technology failure, pointing to slower but deeper workflow-level reengineering as the proven path to production.
Local view
Japan: A press release published 2026-09-26 by ProGuide (via PR Times) presented its "DX・AI, Talent, Back-Office Reality Survey 2026" covering retail and food-service companies, capturing the local bind: companies feel labour shortages acutely and want to use AI, but adoption in back-office operations remains uneven. Separately, Japan Keizai Shimbun (Bunka News) reported on 2026-09-24 a survey of 1,200 IT-purchase decision-makers on how generative AI is changing IT buying behaviour (paywalled). Also, an Impress survey (こどもとIT, 2026-09-25) found roughly 40% of generative AI users in Japan are dissatisfied with "misinformation" from AI tools, and those who fact-check report dissatisfaction most — a quality concern feeding into enterprise trust and adoption hesitancy.
Context & numbers
- IT budgets are set to grow 5.8% in 2026, with 66% of buyers increasing AI spending; high-maturity adopters report 19% ROI vs. 9% for laggards (BCG, ~2 weeks old — confirming figure as context).
- Enterprise AI spending is expected to average $2,068 per employee in 2026, up over 50% from $1,358 in 2025.
- Gartner predicts more than 40% of agentic AI projects will be cancelled by end-2027, even as 40% of enterprise applications are expected to include task-specific agents by end-2026.
- Analytics Insight reports agentic AI in a new phase: 62% experimenting, 23% scaling, with AI spending at $64 billion.
On the radar
- 2027 budget cycles: expect more CFO–CIO friction over AI funding as planning finalizes through Q4 2026 (Innover, Accordion both flag this).
- CDO Vision Tokyo panel takeaways (published 2026-09-24): CIOs from Volkswagen, Carrier and Siemens on scaling enterprise AI — watch for follow-up regional survey data from this event.
- Korea: Jeonju was selected for a KRW 736.8 billion national "physical AI" flagship project, per Seoul Economic Daily (2026-09-28) — a signal for public-sector AI investment lines.
This content was collected, curated, and summarized entirely by AI — including how and what to gather. It may contain inaccuracies. Crew does not guarantee the accuracy of any information presented here. Always verify facts on your own before acting on them. Crew assumes no legal liability for any consequences arising from reliance on this content.
Powered by
