Executive Summary
A new survey of Fortune 500 and Forbes Global 2000 companies, released August 6, found that 74% of enterprises now have AI in production — but around half cannot demonstrate whether it is actually delivering results. At the same time, buyers now rank compliance and explainability above raw performance when choosing AI vendors. The gap between deploying AI and proving its value has become the defining challenge for large businesses heading into the 2027 budget cycle.
Estimated reading time: 7 minutes
Key Takeaways
- 74% of large enterprises have AI in production, but roughly half cannot show whether it is delivering measurable results.
- Compliance and explainability (92%) now matter more to AI buyers than raw performance (74%) — a reversal from prior years.
- Companies with strong data integration report 10.3x return on their AI investment, compared with 3.7x for companies with poor data connectivity.
- Only 5% of surveyed companies consider themselves genuinely “AI native.”
- CEOs, not technology leaders, are now the primary decision-makers on AI strategy at 72% of companies — but formal AI ownership is fragmented.
- The data integration gap, not model choice, is the strongest available predictor of whether an AI investment actually pays off.
What Happened?
On August 6, Plug and Play released its 2026 Enterprise AI Strategy Pulse Survey, covering Fortune 500 and Forbes Global 2000 companies. The survey found that 74% of these large enterprises now have AI systems running in production. However, approximately half of respondents said they could not clearly demonstrate whether those AI systems were delivering a measurable return.
The survey also found a shift in how companies choose AI vendors. Compliance and explainability were cited as a priority by 92% of respondents, ahead of raw performance at 74% and flexibility at 53%. This marks a reversal from the technology-first buying criteria that dominated earlier in the AI adoption cycle.
Only 5% of companies surveyed described themselves as genuinely “AI native.” The survey also found a strong link between data infrastructure and AI returns: companies with strong data integration reported an average 10.3x return on their AI investment, compared with 3.7x for companies with poor data connectivity — nearly a threefold difference.
On AI governance, separate research cited alongside the survey found that 72% of CEOs are now the primary AI decision-maker at their companies, according to BCG. Despite this, formal ownership of AI strategy remains split: only 21% of companies place AI ownership with a dedicated Chief AI Officer or Centre of Excellence, while 37% leave it with individual line-of-business heads.
Why It Matters
For the past two years, the central AI question for large businesses was whether to adopt it. This survey shows that question has already been answered — 74% have. The real question now is whether businesses can prove any of it is working, and most cannot.
This matters because 2026 has been a year of heavy AI investment across nearly every large company. If half of that investment cannot be tied to a measurable result, the businesses heading into 2027 budget planning without clear metrics are at real risk of losing funding, momentum, or both — regardless of how much AI capability they have deployed.
The finding on data integration is arguably the most useful number in the entire survey: a 10.3x versus 3.7x return gap is a concrete, evidence-based argument for where the next round of AI budget should go — not necessarily toward another model or pilot project, but toward the data infrastructure that determines whether AI investments pay off at all.
Business Impact
Enterprise Leadership and CFOs
With CEOs now the primary AI decision-maker at 72% of companies, AI investment decisions are increasingly happening at the top of the organisation — but without the measurement discipline typically applied to other capital investments. Leadership teams should treat AI ROI measurement as a governance requirement, not an optional add-on.
IT and Data Teams
The 10.3x versus 3.7x ROI gap tied to data integration quality suggests that the highest-value AI investment many companies can make right now is in data infrastructure and connectivity, not in acquiring additional AI models or tools.
Procurement and Vendor Management
Vendors should expect buyers to lead with questions about compliance, explainability, and governance rather than benchmark scores. Businesses evaluating AI vendors should update their scorecards to reflect this shift before their next procurement cycle.
AI Governance and Strategy Teams
With AI ownership split between a small number of formal Centres of Excellence (21%) and a larger number of individual business unit heads (37%), many companies likely have inconsistent AI governance across departments. This is a natural area for consolidation.
Key Changes
- Vendor selection criteria have shifted from performance-first to compliance-first, with 92% of buyers prioritising compliance and explainability over the 74% who prioritise performance.
- Data integration quality has emerged as a clear, quantifiable driver of AI ROI, with a nearly threefold difference in returns between strong and weak data connectivity.
- CEOs have become the dominant AI decision-makers at most large companies, shifting AI strategy ownership away from purely technical leadership.
- Formal AI governance structures, such as Chief AI Officer roles or Centres of Excellence, remain the exception rather than the rule.
| Vendor Selection Priority | Share of Enterprise Buyers |
|---|---|
| Compliance / explainability | 92% |
| Performance | 74% |
| Flexibility | 53% |
| Data Integration Quality | Average AI ROI |
|---|---|
| Strong data integration | 10.3x |
| Poor data integration | 3.7x |
Callout — The Real ROI Lever: The single most actionable number in this survey is the 10.3x vs 3.7x ROI gap tied to data integration. Before funding another AI pilot, many businesses would get a better return by fixing the data pipelines feeding their existing AI tools.
Industry Reaction
Plug and Play’s survey, conducted across Fortune 500 and Forbes Global 2000 companies, frames the finding as evidence of a broader “measurement gap” rather than an adoption gap — the technology itself is largely in place, but the tools to evaluate its performance often are not.
BCG’s research on AI decision-making, cited alongside the survey, points to a leadership shift: AI strategy is now predominantly a CEO-level concern rather than one delegated entirely to technology or data leadership, even as formal governance structures lag behind that shift.
Opportunities
- A clear, evidence-based investment case. The 10.3x vs 3.7x ROI differential gives finance and IT leaders concrete data to justify investment in data infrastructure over additional AI tooling.
- Vendor differentiation through compliance. AI vendors that lead with strong compliance and explainability credentials have a clear opening, given that 92% of buyers now prioritise this over performance.
- Governance consolidation. Companies that formalise AI ownership now, ahead of competitors still operating with fragmented, business-unit-led AI strategy, may gain a coordination advantage.
Risks & Limitations
- Unmeasured AI spend is a budget risk. Businesses unable to demonstrate ROI on roughly half their AI investments face real exposure heading into 2027 planning cycles.
- Fragmented ownership creates inconsistent governance. With AI strategy split between a minority of formal Centres of Excellence and a larger share of individual business units, standards and risk controls may vary significantly across the same company.
- “AI native” remains rare. At only 5%, most companies are still adapting existing processes to AI rather than being built around it — a gap that affects how quickly ROI can realistically be expected.
Future Signal Tip: Before requesting more AI budget for 2027, check whether your business can produce a clear before-and-after metric for at least one existing AI deployment. If it cannot, that measurement gap — not a lack of AI tools — is likely the actual constraint on returns.
Future Outlook
Over the next 12 months, it is likely that more large companies will formalise AI governance structures, given the current mismatch between CEO-level decision-making authority and the lack of dedicated ownership roles at most companies. This is an editorial inference based on the direction of the data, not a confirmed trend.
It is confirmed that the survey’s underlying pressure — half of AI deployments lacking demonstrable ROI — will collide directly with 2027 budget planning cycles, making measurement capability a near-term priority rather than a longer-term one.
Vendor messaging is likely to shift further toward compliance and governance credentials over the next 6 to 12 months, following directly from the buyer priorities already reported in this survey.
The Future Signal
The long-term signal in this survey is that AI adoption has outpaced AI measurement, and the businesses that close that gap first will have a real advantage. Deployment was never the hard part — most large companies have already done it. Proving value is the harder, less glamorous work that most have not yet done.
The 10.3x versus 3.7x ROI difference tied to data integration is the clearest evidence yet that the next phase of enterprise AI value will come from infrastructure discipline, not from chasing the newest model. Businesses that treat data plumbing as unglamorous but essential will likely outperform those still funding pilot after pilot without a clear way to measure what any of them are actually returning.
Watch for AI governance to consolidate over the next year, as the current gap between CEO-level ownership and formal accountability structures becomes harder for boards to ignore.
What Businesses Should Do Next
- Test: Pick one existing AI deployment and build a clear before-and-after metric for it before requesting further AI budget.
- Monitor: Track how much of your current AI spend has a demonstrable ROI attached, and flag the portion that does not.
- Prepare: Update AI vendor evaluation criteria to reflect the shift toward compliance and explainability as leading purchase factors.
- Implement: Prioritise data integration and connectivity projects, given the nearly threefold ROI difference tied to data quality.
- Reassess: Review whether AI strategy ownership is clearly assigned within your business, or fragmented across business units without central coordination.
Frequently Asked Questions
What does it mean that half of enterprises “can’t prove ROI” on AI? It means these companies have AI systems running in production but lack the metrics or baselines needed to clearly show whether those systems are saving money, generating revenue, or improving outcomes.
Why does data integration make such a large difference to AI ROI? AI tools generally perform better and deliver more reliable results when they can access clean, well-connected data across a business, rather than working with fragmented or siloed information.
Does this mean businesses should slow down AI adoption? Not necessarily. The survey suggests the priority should shift from adopting more AI tools to measuring and improving the ROI of what is already deployed.
Why are CEOs increasingly the primary AI decision-makers? As AI strategy has grown into a company-wide priority with financial and competitive implications, more CEOs are taking direct ownership rather than delegating the decision entirely to technical teams.
Should our business appoint a Chief AI Officer? Not every business needs a formal Chief AI Officer, but this survey suggests that companies without any centralised AI governance function may face more inconsistent AI strategy across departments.
What should a business measure first to close its own ROI gap? Start with a single, well-defined AI use case and establish a before-and-after baseline, rather than trying to measure ROI across every AI deployment at once.
Is compliance really more important than performance when choosing AI vendors now? According to this survey, yes — 92% of buyers now prioritise compliance and explainability over the 74% who prioritise performance, a meaningful shift from earlier AI buying patterns.
Conclusion
The headline finding here is not that AI adoption has stalled — it clearly has not, with nearly three-quarters of large enterprises running AI in production. The real story is that measurement has not kept pace with deployment, leaving half of that investment unaccounted for. The businesses that close this gap by 2027 — starting with data integration and clear ROI baselines — are likely to pull ahead of competitors still funding AI deployment without proof it works.
