AI Reality Check: Going For Growth

This voice experience is generated by AI. Learn more.This voice experience is generated by AI. Learn more.Dan Priest, U.S. Chief AI Officer at PwC.

getty​Roy Amara famously said that “we tend to overestimate the impact of technology in the short run and underestimate it in the long run.” In my opinion, that’s exactly what’s happening with AI.

The bigger story is what AI gives back. As the technology absorbs more work, it returns time, expertise and capacity to the business—a workforce dividend that most organizations haven’t yet decided how to spend. The ones pulling ahead are going for growth, and that decision may prove more consequential than the technology itself.

What’s different this time is the pace. According to a November 2025 Federal Reserve Bank of St. Louis report, GenAI adoption has occurred faster than both PCs and the internet at comparable stages following their mass-market introduction. That compressed adoption curve means delays in decision making can create gaps that widen by the month rather than the year.

I believe the truth of this moment is that AI success will be about 20% technology and 80% transformation. The technical capability is impressive, but the real value comes from changing how people, processes and organizations adapt around it. AI should be intrinsic to business strategy, not treated purely as a technology play.

AI has made remarkable progress over the past two years. Systems are more reliable, their reasoning is more consistent, and their ability to stay on task has improved dramatically. Today’s systems can sustain focus across longer and more complex chains of reasoning, making them practical co-creators for research, analysis and decision support.

Better technology doesn’t automatically mean better economics, though. Our research has found that token prices keep falling while AI bills keep climbing, which is why many CIOs and CFOs are focused on cost discipline: underwriting use cases before they’re built, routing tasks to the most cost-effective model and measuring cost per outcome instead of usage. When everyone is running the same models, how much AI you get per dollar becomes a differentiator the models themselves can’t provide.

The technology isn’t perfect, but the progress is real. While it’s tempting to fixate on the next leap in capability, I’d argue that consistency is the real breakthrough. The magic of AI in 2026 will be that it delivers predictable, repeatable outcomes. That reliability, priced with discipline, is what makes enterprise-wide transformation possible.

Growth plays should be funded at a level that actually moves the business, which turns the question into a sizing question: Are we deploying enough AI—and in the right way—to actually change performance?

That question has a threshold. Our benchmarking analysis found that when companies invest below 1.6% of revenue into AI, returns tend to stay insignificant. When they invest above 1.6% of revenue, outcomes surge: EBITDA is up 9.5%, total shareholder return is up 20.2%, and revenue is up 3.5% relative to peers. Underinvestment doesn’t buy a smaller version of the same return; it often buys nothing at all.

Initially, activity often measured AI success (pilots launched, models deployed, tools licensed), but those metrics don’t tell us whether AI is delivering lasting value. Organizations pulling ahead are shifting toward outcome-based benchmarks that track productivity, cycle time, quality and decision velocity. They’re measuring impact, not just adoption.

Benchmarks also help leaders manage AI as a portfolio rather than isolated use cases. The strongest performers are running a bimodal transformation—concentrated reinvention in a handful of high-value domains paired with scaled enablement across everything else. PwC found that fewer than 5% of S&P 500 companies are doing this today, and those that do report twice the revenue of their peers.​

Above all, benchmarks give leaders a shared definition of success, making it easier to prioritize investments, course-correct quickly and scale what works. That clarity is a strategic advantage in a rapidly accelerating AI landscape.

The technology may feel complex, but the hard part is still the human element––the organization that changes around it.

The companies getting the most out of AI are the ones rethinking how people, processes and technology come together. That work starts with a choice about capacity. AI frees up human time and expertise, creating a workforce dividend that can either be banked as margin or reinvested into new products, markets and roles. Banking the savings can give you a gain your competitors can match once they adopt the same tools. Reinvesting builds capabilities they can’t easily copy.

You can see that reinvestment in how the leading companies operate. They’re applying their best talent to the toughest challenges, training leaders to use AI responsibly and giving teams permission to experiment safely.

In every conversation we have with leaders, one theme comes up: Innovation is born of tech and talent. That’s why the game changers in this next chapter will be the people who know how to orchestrate the agents—who bring curiosity, judgment and creativity to the table and let AI amplify those strengths.

If Amara’s Law holds true, we’re now underestimating what comes next. We already know the technology works.

The next phase will be about what we choose to build with the capacity AI creates, with the trust and governance to make it stick. Benchmarks ground that ambition in evidence and keep the dividend pointed at growth rather than quietly absorbed into savings.

This also requires rethinking how teams collaborate with tech and redesigning roles, workflows and incentives to reflect a future of co-working with intelligence.

Most importantly, it means focusing on outcomes that matter, not outputs that impress. The organizations that succeed will be the ones balancing innovation with discipline and building for reliability, scalability and human-centered impact.

Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?