The AI Productivity J-Curve: Why Companies Aren’t Seeing ROI Yet | Chief AI Officer Podcast

Sanjay Puri (0:00)

Sanjay Puri, Chief AI Officer Connect Podcast Economists describe new technology adoption with something called a productivity checker. In the early years, companies spend heavily buying chips, buying tools, investing in data centers, retraining people, redesigning how work gets done before they see much payoff, if any. Costs go up, and measured productivity can actually dip or flatten before it eventually starts up sharply.

Once the investment starts paying off and plotted on a chart, that dip then rise looks exactly like the letter J. And you'll hear it referenced today as a shorthand for, we've paid the upfront cost, and we are now waiting on the payoff. The real debate is going to be how far along that curve companies actually are. So let's get into it. Where do we stand now at mid-2026? Now, this episode also doubles as a half-year review. And before we get into specific debates, one round of quick, direct answers from esteemed panelists. You know, where does the typical large enterprise actually stand on AI right now? Six months, a little more than seven months into it. Eric, we'll start with you.

In plain terms, is the typical Fortune 500 or Fortune 1000 company ahead of where you'd expect them to be in 2026, on track or behind? What would you say?

Eric (1:40)

I've never seen such a big gap between the technology's potential and what's actually being delivered on the ground. I see this as a huge opportunity.

I have to say, I'm not that surprised, because I've been studying technology for basically all my career, and this happens every single time. You know, if you go back to the steam engine or electricity, these were momentous technologies that completely transformed how the economy worked. And a lot of people looked at just the core capabilities and were like, wow, Andy and I wrote a book about this.

But in each case, getting management to take full advantage of it, re-skilling the workforce, changing the business processes, just understanding where the real opportunities are, that always takes a lot longer. Quantitatively, there's 10 times as much effort required to implement these technologies as there is to discover them. Most people don't see that because it's intangible. It's a big part of why we created Work Helix, to speed up that gap. So yes, it's taking a while. It's faster than it was with earlier technologies, and with our help, it's going to be faster yet.

Sanjay Puri (2:44)

Andy, coming to you, from what you're seeing from your perspective, what's the single biggest change in how companies are using AI right now? To let's just say, maybe six, seven months before. Do you see any change at all?

Andy (2:59)

Well, Sanjay, there's one gigantic change, and you picked the right starting point, because the improvements in the models and the harnesses from January forward have been stupefying. Like Eric says, he and I have been studying technology for a long time. I've never seen anything that made me think I was living in science fiction, and I've never seen a technology prior to AI that improved as quickly. Digital technologies improve very, very quickly by the standards of anything else in the economy. But AI is just leaving everything else behind. So the big change in 2026 is these models were already good before. Now, they're just kind of spectacular, jaw-droppingly good.

And enterprises are taking advantage of that, but they're doing it in a really uneven way. There are pockets of enthusiasm about what models can do and what the latest capabilities are, and they're concentrated in parts of the organization that you would expect, like software engineering. But so far, there's still pockets. And one of the huge opportunities that companies have is to democratize not just access to the technologies, but use of the technologies, and taking advantage of more of the amazing capabilities that the latest models have, and making sure people are aware of and have the opportunity to learn and experiment with them.

Sanjay Puri (4:18)

James, you work inside company accounts sometimes, and you work with them directly.

What's the one thing that has surprised you? Andy talked about the models becoming better, greater, the adoption, et cetera. But from a client standpoint, is there one thing that has surprised you in the first half of 2026?

James (4:39)

Being able to numerically quantify the gap that Eric and Andy are speaking about. The executives that I speak to, when we show them the data, they go, wow, we are clearly on a track to capture 1 percent of this opportunity.

And like Eric has said here, that history continues to repeat itself. CRM, cloud, the Internet, this is no different. The technology itself, it does not magically change a company. There's an immense set of complementary and other investments and change management and measurement that needs to get put in place to even begin to unlock this technology. So even though in 2022 is when a lot of this made a big splash, we are years later and still on track to capture about 1 percent of the opportunity.