If there was such a thing as the 2025 AI Quote of the Year, MIT Sloan would win, hands down, for their statement that “95% of enterprise AI initiatives are failing to deliver a return on investment.” To any serious observer of the most consequential phenomenon of our time, the problem isn’t the technology. It’s how we measure it. In reality, the most successful companies are not chasing “project payback” but are measuring “capacity unlocked”.

As we enter 2026, one thing is clear: AI is no longer a speculative topic or a future promise. It is quietly reshaping how companies operate, scale, and are valued.

The observations below are drawn from executive conversations, client work, and recent research, inspired by a grounded Wall Street Journal CIO Journal roundtable, alongside publications from McKinsey, Harvard Business Review, Microsoft, and OpenAI.

Below are my takeaways, ordered intentionally as a countdown, from the most visible effects of AI adoption to the deeper operating insight that ultimately explains the rest.

 

My Countdown

 

#9. AI has become a value-creation signal, not an IT initiative

AI maturity is increasingly read by investors and boards as a proxy for management quality, scalability, and execution discipline, particularly in PE-backed and exit-minded companies. (Source: Harvard Business Review – How AI creates real business value)

 

#8. Competitive advantage is shifting from models to people

Across companies, the bottleneck is no longer access to technology. It is adoption, permission, and curiosity. Tools are widely available; meaningful usage is not. (Source: OpenAI – The State of Enterprise AI)

 

#7. Fear of vendor lock-in is shaping AI architecture decisions

Organizations are deliberately staying model-agnostic, informed by hard lessons from earlier cloud lock-in cycles. (Source: Wall Street Journal – Tech Cloud and Vendor Lock-in)

 

#6. AI is reshaping hiring quietly, not dramatically

AI investment increasingly coincides with slower hiring, not mass layoffs. Most firms are simply absorbing growth differently. In many cases, they are choosing not to backfill roles because AI has absorbed the workload. (Source: Wall Street Journal – AI and Hiring Trends)

 

#5. Customer-facing AI was delayed by trust, but momentum is building

Concerns around hallucinations, brand risk, and compliance slowed adoption. As guardrails improve, hesitation is giving way to controlled experimentation. (Source: Harvard Business Review – Why most AI pilots fail)

 

#4. Agentic AI is real, but governance is the constraint

AI systems that take action are entering production, but autonomy remains intentionally limited by leadership caution. (Source: Wall Street Journal – AI Agents in the Enterprise)

 

#3. Copilots became table stakes; differentiation moved elsewhere

Rolling out copilots is now baseline adoption. Advantage comes from redefining role-level outputs, not distributing tools. (Source: Microsoft – Future of Work Report 2025)

 

#2. AI is delivering value, just not where the hype predicted

Most companies are using AI daily for workflow automation, summarization, and research. The gains are incremental but real. (Source: WSJ CIO Journal – What companies are actually doing with AI)

 

…and the deeper operating insight that ultimately explains the rest:

#1. ROI Was Reframed: From Project Payback to Capacity Unlocked

Executives did not abandon financial discipline in 2025, nor did they stop demanding returns from AI investments. What changed was the framing of the discussion.

Midway through the year, a widely cited MIT Sloan study triggered controversy by suggesting that more than 95 percent of enterprise AI initiatives were failing to deliver their expected return. While the headline energized AI skeptics, its more important effect was to force a deeper examination of how, where, and over what time horizon AI actually creates value inside organizations.

The debate exposed a structural mismatch. Many AI initiatives labelled as “failures” were evaluated as discrete projects, expected to justify themselves through traditional, one-off payback logic. In practice, that approach misses how AI delivers economic impact.

What became clearer through 2025 is that AI adoption behaves less like a project and more like an operating capability. Its benefits rarely materialize as a single, visible win. Instead, they accumulate through capacity unlocked across the organization: time released from manual work, bottlenecks removed from core workflows, and hiring avoided or deferred as existing teams are able to handle greater complexity and volume.

In companies seeing tangible results, that capacity is not left idle. It is steadily reallocated away from low-value, legacy ways of working and toward activities that support growth, responsiveness, and scale. This is why many of the most effective AI initiatives appear modest on the surface. They are narrowly scoped, quick to implement, and deliberately non-disruptive, yet they often pay for themselves rapidly, frequently within 30 to 60 days, by expanding what the organization can do without expanding headcount.

Seen this way, AI adoption closely resembles earlier technology inflection points. In the early smartphone era, most applications were tried and discarded. A small number stuck, quietly changing habits and expectations. Over time, those small changes compounded into fundamentally new ways of working. AI is now following a similar pattern inside companies.

For mid-market organizations, this reframing is decisive. The winners in 2026 will not be those pursuing large, multi-year AI programs, but those that treat AI as a continuous discipline, systematically reallocating effort and spend away from stagnant processes and into higher-value work, while building the organizational muscle to experiment, learn, and compound gains without destabilizing operations.

(Sources: MIT Sloan research on enterprise AI ROI (reported by Healthcare IT News), and McKinsey & Company, The State of AI 2025.)

 

So, What now?

Which of these nine observations reflect what you are seeing on the ground? Are you experiencing capacity unlocked in your workflows? Or still chasing project-by-project payback?

2025 showed that AI is neither a silver bullet nor a disappointment. It is an amplifier.

How powerful an amplifier? One useful macro signal comes from JP Morgan, which estimates that since the launch of ChatGPT in late 2022, AI-related companies have accounted for roughly 75 percent of S&P 500 total returns, 80 percent of earnings growth, and 90 percent of capital-expenditure growth. While attribution can be debated, the conclusion is hard to ignore: value creation has been overwhelmingly concentrated in AI-linked companies.

For mid-market organizations, the 2026 challenge is not to lead with ambition, but to remove friction, unlock capacity, and strengthen decision-making before automating everything. As AI lowers barriers to access markets, ecosystems, and capabilities once reserved for much larger players, scale advantages are eroding. Competitive advantage is shifting toward scope, an arena where focused, agile mid-market companies can excel.

The question is no longer whether AI delivers value, but how deliberately organizations convert unlocked capacity into growth.

At Newport LLC, my partners and I work with mid-market leaders who intend to use this window to pull ahead, not keep pace. If you want to pressure-test where AI is truly creating advantage in your business, and where it isn’t, we would welcome the conversation.