Welcome to the HBR Executive Agenda for August 28, 2025. In This Issue: My AI Investments Haven’t Paid Off. Should I Tap the Brakes? A High-Impact Strategy Exercise for Your Next Leadership Offsite My AI Investments Haven’t Paid Off. Should I Tap the Brakes? A well-publicized new MIT study found that “95% of organizations are getting zero return” so far from the billions they’ve collectively invested in GenAI. The report is generating questions among executives everywhere. Is this just a normal part of the development curve? Or have we over-invested in what’s still a largely unproven technology? There’s no clear answer to these questions, at least not yet. But executives are hungry for some sense of direction, so I reached out to several experts to get their insights on the state of AI and on how businesses should think about how they invest in the technology. Tom Davenport, Babson College professor and research fellow at the MIT Initiative on the Digital Economy, thinks we’re in a bubble. It’s pretty clear that bottom-up experimentation with generative AI in large organizations isn’t yielding much measurable value, which will lower demand. Enterprise-level use cases take substantially longer to implement, which will also probably lower demand. For some use cases, such as code generation, the productivity gains apparently are not quite what was anticipated; that should lower demand too. And some data suggest that the general public doesn’t use gen AI tools very often. All of this suggests that there is a mismatch between corporate and individual demand vs. the amount being invested in LLM providers, new data centers, etc. I think that is suggestive of a bubble that will pop or deflate soon. Amy Webb, CEO of the Future Today Strategy Group, says companies are investing in AI without a clear strategy. Every transformative technology creates a hype cycle. The question isn’t whether AI will reshape business—because it has and will continue to do so—but whether your current investments will survive the inevitable correction. AI is a marathon technology in a sprint-funding world. The mismatch between long-horizon development timelines and immediate investor expectations is where bubbles are born. What I’m seeing in many businesses is a willingness to spend millions on solutions before fully defining the problems AI is intended to solve. Or worse, some businesses are making quick millions by selling entire archives before defining a long-term business model. Leaders are making decisions out of fear or the fear of missing out, rather than using strategic foresight. History shows us that during tech bubbles, fortune favors companies that focus on fundamentals. But I don’t see enough companies thinking about either the fundamentals or the next-order implications of their decisions. AI might be cheap at the start, but the costs to scale and continually train data can be far beyond what one’s finance team is imagining. The real AI winners won’t be the first movers or the biggest spenders. They’ll be the disciplined operators who implement AI where it creates measurable competitive advantage. But you must do the hard, data-driven work to model future scenarios to identify where the world is going, where value will be created, and how your organization will participate in the AI era. Chris Kempczinski, CEO of McDonald’s, which has invested in various AI experiments to modernize its business. I think AI is a game-changer, but it will take longer than most industry prognosticators expect to see results. Most CEOs would rather be accused of going in too aggressively than being asleep at the switch, particularly if this is a technology that could disrupt your industry. We need to be there, exploring use cases. The investment required isn’t material. But the payoff, if it works, could be massive. As we experiment with use cases, the biggest challenge is stitching together the disparate data we already have so that AI can work with it. This will take time. Tech CEOs have told us they’re not worried about AI’s capability but with getting it to work with clients’ legacy data systems. Mark Mortensen, INSEAD professor, urges companies to think about costs and benefits beyond immediate ROI. It’s easy to get excited about what AI can do, but it’s challenging to translate that into something viable. Yes, there’s some low-hanging fruit for successful gen AI applications: fact checking, creating PowerPoints, some coding. But it will take time and learning to figure out how to successfully integrate AI into business planning and strategy. I would push companies to think hard about the human side of what’s happening. That will require an audit that looks at more than just immediate ROI. For example: If AI is replacing entry-level work, then how will we train our new hires? We need to think hard about the long-terms costs and benefits of new forms of training, learning, and social connection. Randy Bean, author and consultant, believes business leaders need to view AI from a long-term perspective. AI will transform our world, but we are still at the very early stages of the journey. Like technology transformations throughout history, the impact is often overestimated in the short term and vastly underestimated in the long term. Business leaders need to step back and envision how AI can transform their business and their industry over the coming decades. The race doesn’t always go to the first out of the gate but to those who develop a vision that they execute on relentlessly but patiently. Dan Priest, chief AI officer, PwC US, isn’t convinced that we’re even in a bubble. Gen AI has sparked rapid adoption and learning across industries. While not every early initiative has delivered lasting impact, organizations are building on these lessons and increasingly achieving meaningful, scaled benefits. These successes are strengthening confidence and encouraging continued investment. It’s still early to predict the long-term trajectory and the possibility of an AI bubble, but momentum is still strong. No one is doubting the power of gen AI to eventually upend the business world. The question is about timing. Businesses need to keep experimenting but should do so with purpose and in pursuit of clear strategic objectives. Winning with AI isn’t a sprint, but a marathon. A High-Impact Strategy Exercise for Your Next Leadership Offsite By Ann Hiatt Each September, my calendar fills up with strategy summits I’m facilitating for clients looking to assess, pivot, and reset to hit their yearly goals. My new CEO clients often come to me dreading these sessions. They complain their teams cram in too much content, avoid real tension, or treat offsites like a corporate vacation. To produce an effective strategy offsite, you need to keep it simple, focused, and consistent. You want just enough structure to get the team to honest conversation and clear decisions. To set the tone, I recommend an exercise I’ve conducted with dozens of clients. It’s called PechaKucha—named after the Japanese word for “chit chat”—and it’s a round of fast-paced, concise, visual presentations. It’s my secret weapon for keeping things moving, forcing clarity, and surfacing essential themes for deeper discussion. How It Works About a week before your offsite, ask each member of the executive leadership team to create a 10-slide presentation that addresses the following: Their proudest moment of the year (so far) What’s working What isn’t working Their biggest blockers or challenges Their biggest opportunities Key metrics that indicate what’s going well and what needs attention Their top lesson(s) learned Their vision for the next year What they need to make their vision happen Their HERO (highly engaged & resourceful operative), aka a high-performing member of their team Executives should be told in advance that they’ll have 20 seconds to present each slide—which means each 10-slide presentation is limited to 3 minutes and 20 seconds max! Coach the presenters to stay focused on one image or idea per slide, because in the room the slides will be set to automatically advance after 20 seconds. I’ve seen really impactful results from this seemingly simple exercise. One client recognized that a bureaucratic core process was throttling progress, and he put an immediate stop to it. Another time, the newest member of my client’s ELT had an observation that led to an entirely new product offering that revitalized previously stagnated sales. Both clients told me afterwards that they were sure these moments of unlock never would have happened without this rapid-fire exercise. The PechaKucha format works for the following reasons: It forces clarity. The time limit means leaders only share what they really think matters. It energizes the room. The rapid pace keeps people engaged and responsive. It levels the field. Everyone gets the same amount of time—no hierarchies, no fluff. It surfaces priorities. You’ll see patterns in blockers, opportunities, and needs across the org. The themes and collaboration opportunities you identify from everyone’s presentations will inform the rest of your deeper strategic work over the course of the offsite. Ann Hiatt is an executive leadership consultant and author of the 2021 book Bet On Yourself.
