Two massive new studies dropped this week — and they confirm what too many enterprise leaders suspect but won't say out loud: most AI programs are failing to deliver measurable value, and the gap between winners and everyone else is widening fast.
Info-Tech Research Group surveyed 551 senior leaders in June 2026 and found that enterprises with a dedicated, governed AI strategy are three times more likely to report measurable AI impact — 60% versus just 20% for organizations with no active strategy. That same week, SAP and Oxford Economics released their Value of AI Report 2026, surveying 2,600 business leaders across 13 countries, confirming that while global AI ROI is rising, the distribution is deeply uneven.
The data isn't abstract. It's a clear diagnosis of what separates the enterprises realizing AI value from the ones still running pilots with nothing to show for it.
The Uncomfortable Number: 20%
If your organization doesn't have a formal, board-governed AI strategy, there's an 80% chance you're not seeing measurable AI impact. That's not a pessimistic spin — it's what the Info-Tech data shows directly.
The 60/20 split deserves a moment. Sixty percent of organizations with a formal AI strategy report measurable impact. Twenty percent without one do. This is a 3x multiplier — and unlike most enterprise technology gaps, it's not driven by budget. In fact, the organizations spending the most on AI aren't automatically the ones seeing results.
The SAP/Oxford report puts a dollar number on this. The average global company is now spending $28 million annually on AI. Expected ROI sits at 21% — about $6.3 million back on that investment — up from 16% last year. That sounds promising. But dig into the data and you find that 69% of businesses are "satisfied" with their AI ROI, even though the majority don't believe AI is operating anywhere near its full potential.
Satisfied with underperformance is a dangerous place to be.
What "Formal Strategy" Actually Means
Here's where most executive conversations go wrong. When people say "formal AI strategy," they picture slide decks, steering committees, and governance theater. The Info-Tech data points to something more concrete.
Brian Jackson, principal research director at Info-Tech, put it plainly: "The organizations seeing measurable impact are not treating AI as a collection of disconnected use cases. They are connecting AI to strategy, data readiness, executive accountability, and clear measures of business outcomes."
Translated: a formal strategy has four non-negotiable components.
Data readiness comes first. Organizations achieving department-wide AI adoption with measurable impact rate their data quality as excellent. This isn't a coincidence — it's causation. You can run the best model in the world on bad data and get garbage. The SAP report confirms this independently: 73% of global companies report challenges with incomplete data, and 79% experience rework, delays, or backlogs due to low-quality AI outputs. Data quality isn't a technology problem. It's a governance decision that happens before the first AI model goes into production.
Executive ownership is required, not optional. The Info-Tech data shows CIOs and CTOs remain the primary AI owners in most organizations — leading AI in more than half the companies surveyed. That's fine as a starting point. But here's the finding that matters: organizations with dedicated Chief AI Officers currently report the highest rate of department-wide adoption with measurable impact. Not because the CAIO title is magic, but because having a single executive who owns AI outcomes — accountable to the board, not buried inside IT — changes the incentive structure.
Board-level governance unlocks budget confidence. Info-Tech found that 73% of organizations with a formal, board-governed AI strategy report high confidence in AI budget increases over the next 12 months. Among organizations with only ad hoc or department-led strategies, that number drops to 34%. You can't build enterprise-scale AI on uncertain budget cycles. Governance isn't bureaucracy — it's the mechanism that keeps funding stable.
Measuring the right outcomes is everything. This is the one that most enterprises get badly wrong.
The Cost Reduction Trap
Among the most impactful AI use cases identified in the Info-Tech study, only 11% cite cost reduction as the primary goal.
Read that again.
The boardroom pressure on AI usually runs in one direction: reduce costs, cut headcount, show efficiency gains. But the organizations actually delivering measurable AI impact are not primarily targeting cost reduction. They're targeting productivity and throughput (38%), revenue growth, risk reduction, quality and accuracy, customer satisfaction, and regulatory/compliance outcomes.
Cost reduction as the north star for AI is a trap. It creates programs optimized for short-term savings — automating a task here, cutting a role there — while missing the compounding strategic advantage that comes from using AI to generate revenue, reduce risk, and improve quality at scale.
The CFO reading this has a decision to make: keep measuring AI's value in terms of cost avoided, or start measuring it in terms of strategic outcomes delivered. The companies hitting 60% measurable impact are overwhelmingly in the second camp.
The Buy-vs-Build Reality
The Info-Tech study also clarifies something that was often debated but is now becoming settled: 80% of enterprises prefer to buy AI solutions rather than build them in-house.
The breakdown: 42% activate AI through existing vendors — Microsoft Copilot, Salesforce Einstein, ServiceNow, SAP AI features embedded in software they're already paying for. Another 38% are selecting new, best-of-breed AI vendors. Only 20% are building AI models internally.
This is a significant strategic data point for CIOs and CTOs who are still allocating headcount and compute budget to build proprietary models for use cases where off-the-shelf solutions are already available and battle-tested at enterprise scale.
Building in-house still makes sense in specific scenarios: highly proprietary data, unique competitive requirements, or areas where no vendor solution meets compliance standards. But for the majority of enterprise AI use cases — productivity, knowledge management, code assistance, customer service, process automation — the build-vs-buy math increasingly favors buying and customizing.
And the pressure to make that decision well is intensifying. The Info-Tech data shows 78% of IT executives expect AI to disrupt their current SaaS model within two years. When your existing SaaS vendors are embedding AI into tools you're already paying for, the decision framework changes. Platform AI versus point solutions versus custom builds — all require a formal strategy to evaluate consistently.
The ROI Math Is Getting Harder to Ignore
The SAP/Oxford data on expected ROI has a trajectory that every CFO should be tracking.
Current state: $28M average AI spend, 21% ROI ($6.3M), up from 16% last year. Projected two-year state: 38% ROI ($15.9M). That's a 2.5x increase in ROI on a spend base that's growing modestly.
The agentic AI numbers are even more striking. Expected ROI from agentic AI in the next two years: $17.6 million — more than four times last year's $4.3 million estimate. Organizations deploying AI agents are already reporting average ROI of 171%. U.S. companies specifically are seeing 192%.
The math creates urgency. Companies that formalize their AI strategy now — governance, data readiness, executive accountability — are positioning to capture returns in the 38-171% range. Companies still running disconnected pilots are positioning to remain in the 20% who see no measurable impact.
The Agentic Gap Is Already Opening
The SAP data includes a warning that every enterprise board should be reading carefully: 69% of businesses are deploying AI agents faster than they can govern them.
That's not a technology problem. That's a governance failure happening in real time.
The specifics are striking: 38% have no human-in-the-loop process for agentic workflows. 37% don't have permission and access controls for agents. Only 44% have a registry of the agents running inside their organization. Only 3% consider themselves fully prepared for agentic AI — while 83% believe it has moderate-to-very-high transformation potential.
This creates a dangerous gap between strategic ambition and operational readiness. Agents that act autonomously on business processes — scheduling, procurement, customer communications, financial approvals — without governance frameworks create risk that compounds faster than manual processes.
The organizations with formal AI strategies are addressing this now. They're building agent registries, establishing human-review requirements for high-stakes decisions, and setting permission controls before deployment, not after incidents.
What Leaders Should Do This Quarter
The data from both studies converges on the same prescriptions. Here's the practical translation:
If you're a CIO or CTO: Your first priority isn't model selection or infrastructure. It's data readiness audit and governance framework. Neither requires a budget increase — they require executive time and organizational discipline. The organizations seeing 60% measurable impact built their data foundation before scaling AI programs. Start there.
If you're a CFO or COO: Reframe how you measure AI value. Cost reduction as the primary metric is holding back your organization's AI program. Expand measurement to include productivity throughput, revenue impact, risk reduction, and quality outcomes. Demand ROI reporting that captures all four dimensions, not just headcount equivalent.
If you're a CEO or board member: Two things need to happen at the governance level. First, decide who owns AI — CIO/CTO or a dedicated CAIO — and give that person explicit board-level accountability. Second, build an agent registry now, before you have dozens of autonomous AI processes running without visibility or control. The 69% who are deploying faster than governing will face an incident within 18 months.
For everyone: Consolidate your AI vendor relationship to existing platforms where possible. The 80% buy rate isn't laziness — it's smart resource allocation. The AI your workforce will actually adopt is the AI embedded in tools they already use.
The Window Is Narrowing
Forty-two percent of organizations have achieved department-wide AI adoption with measurable impact, per Info-Tech. That number was nowhere near 42% two years ago. The enterprises in that 42% are building compounding advantages — better data, faster automation, higher workforce productivity — while the other 58% are still in pilot mode.
The Info-Tech finding that 96% of IT executives expect AI budgets to increase over the next 12 months (with 46% expecting increases greater than 25%) tells you the investment phase isn't slowing down. It's accelerating.
The question isn't whether to invest more in AI. It's whether the organizational infrastructure — strategy, governance, data readiness, executive accountability — is in place to convert that investment into the 3x measurable impact difference the data shows is possible.
The strategy gap is real, it's measurable, and it's widening. The organizations on the right side of it are not smarter or better funded. They're more deliberate.
Sources: Info-Tech Research Group AI Adoption and Impact Study: AI in the Enterprise June 2026 (n=551 senior leaders); SAP Value of AI Report 2026 with Oxford Economics (n=2,600 business leaders, 13 countries)
