In January 2025, BCG published its AI Radar report, a survey of more than 1,800 C-suite executives across 19 industries and 100 countries. The central finding was not about which AI tools performed best or which industries adopted fastest. It was about how many bets companies placed.
The top-performing companies in the survey focus on an average of 3.5 AI use cases. The rest spread across 6.1. The focused group generates 2.1x higher ROI.
That single finding would be interesting on its own. What makes it significant is that four additional Tier 1 sources, published independently across 2025, arrived at the same conclusion from entirely different angles. The convergence is unusually clean: the data, the market signals, the analyst predictions, and the financial pressure all point in one direction.
The Data: Focus Beats Spread
BCG’s survey divided companies into three tiers. At the top, 5% qualify as what they call “future-built.” These companies concentrate investment, redesign workflows around fewer AI initiatives, and scale what works before adding anything new. The result is 2x revenue growth and 3.6x three-year total shareholder return compared to the 60% still spreading thin. A follow-up report in September 2025 confirmed the gap is widening, not closing.
McKinsey’s State of AI survey tells the same story from a different angle. Only 39% of organizations attribute any EBIT impact to AI, a finding consistent with the broader measurement gap across the S&P 500. Most of those report less than 5%. Just 6% qualify as high performers, the companies seeing 5% or more EBIT impact. What separates the 6% is not their technology stack or their AI budget. It is that they push for transformative innovation rather than incremental automation, they redesign workflows, and they scale fewer things faster.
The pattern across both surveys is consistent. The companies extracting real value from AI did not get there by buying more tools. They got there by choosing fewer tools and going deeper with each one.
The Market Signal: More Money, Fewer Contracts
If the data describes what works, the market is beginning to enforce it.
TechCrunch surveyed 24 enterprise-focused VCs in December 2025. The consensus was direct: AI budgets will grow in 2026, but the number of vendor contracts will shrink. “Enterprises will cut out experimentation budget, rationalize overlapping tools, and deploy savings into the AI technologies that have delivered,” said Andrew Ferguson of Databricks Ventures. Rob Biederman of Asymmetric Capital Partners was blunter: budgets will increase for a narrow set of products that clearly deliver results, and will decline sharply for everything else.
The language across the 24 responses was consistent. CIOs reducing SaaS sprawl. Unified systems replacing fragmented stacks. Measurable ROI as the new procurement filter. The experimentation window that defined 2023 and 2024 is closing.
Gartner reinforced this in October 2025 with a structural warning: agentic AI supply already exceeds demand, and a market correction is coming. The comparison they drew was to dot-com, telecom, and energy corrections. The technology is sound, they argued, but the number of vendors selling undifferentiated versions of it is not sustainable. Capital-rich incumbents that can acquire promising technology and talent will win. Everyone else will consolidate or disappear.
The CFO Gate
The market correction is not only coming from VCs and analysts. It is coming from inside the building.
Forrester’s 2026 Predictions report found that enterprises will defer 25% of planned AI spending to 2027. The reason: fewer than one-third of decision-makers can tie AI value to financial growth. CFOs are now gatekeeping AI budgets with a question that should have been asked two years ago: which of these tools actually produce measurable results? The tension between AI ambition and financial accountability is not new, but the data is now sharp enough to force a resolution.
This is the mechanism that connects the data to the market signal. When BCG shows that focused companies outperform, and McKinsey shows that 94% of companies are not getting meaningful value, and Forrester shows that CFOs are pulling the purse strings, the path forward becomes narrow. Companies will not voluntarily consolidate their AI stacks out of strategic wisdom. They will consolidate because the CFO stopped approving renewals for tools that cannot prove their value.
What Consolidation Looks Like in Practice
The instinct for most organizations is still to add. Another pilot, another vendor evaluation, another proof of concept. The data from all five sources suggests the opposite.
The companies in BCG’s top 5% did three things consistently. They concentrated investment on a small number of use cases. They redesigned workflows around those use cases rather than layering AI on top of existing processes. And they scaled what worked before evaluating what might work next.
For a CEO reviewing the AI stack, the diagnostic is straightforward. Count the number of active AI tools and pilots. If the number is closer to 6 than to 3, the data suggests you are in the majority that gets half the ROI. The prescription is not to find a better tool. It is to audit what you have, cut what cannot demonstrate measurable results, and redirect that budget into the initiatives that can. The model decision framework matters less than the discipline to pick a number and hold it.
Consolidation is not a retreat from AI ambition. It is the first real signal that a company knows which AI bets actually pay off, and is willing to stop funding the ones that don’t.