Skip to content

Zero Payoff Is Not an AI Problem. It Is a Work Problem.

Zero financial payoff from AI

Zero Payoff Is Not an AI Problem. It Is a Work Problem.

On October 24, Torsten Slok, the chief economist at Apollo, put a number on what many executives have been feeling and few will say out loud. Despite massive investment and hype, he wrote, AI “has delivered zero financial payoff to corporate America outside of tech.” Revenue per worker and profit margins are rising for the Magnificent Seven and flat for what he calls the S&P 493, the rest of the S&P 500. (Apollo, Oct. 24, 2026)

He added that “mainstream corporate America is pulling back and starting to view token spending not as a source of growth but as a cost center that weighs on margins.”

Not one study. A pattern.

Slok is not alone. Survey after survey shows the same thing:

  • NBER, summer 2026. Nearly 9,000 CEOs, CFOs, and senior executives across seven countries: 78% of firms use AI, yet 83% of executives report no impact of AI on their firm’s productivity over the last three years. (NBER Working Paper 34836)
  • BCG, September 2025. In a global survey of 1,250 senior executives, 60% of companies “report minimal revenue and cost gains” from AI. Only 5% are generating value at scale. (BCG, The Widening AI Value Gap)
  • S&P Global Market Intelligence, 2025. The share of companies abandoning most of their AI initiatives before production jumped from 17% to 42% in one year, and the average organization scrapped 46% of its proofs of concept. (CIO Dive)
  • MIT NANDA, 2025. 95% of organizations studied were getting no measurable P&L return from generative AI, a result the authors said “seems to be determined by approach,” not model quality. (The GenAI Divide)

Different researchers, different methods, same picture. The models keep getting better. The results inside most companies don’t. Amanda Luther, a BCG coauthor, named the cause: companies “urgently need to redesign how work gets done.”

A deer in the headlights

I don’t need surveys to see this. I see it directly in the many organizations I work with.

Recently I spoke with a senior scientist at a major research university. We talked about a widespread PHP vulnerability. The scientist told me they don’t know what to do about it, because they don’t even know how many vulnerable websites they have. They are like a deer trapped in the headlights of the oncoming AI.

This is not new. Years ago I held a large consulting contract with a university IT organization. Management said they were doing Scrum. I documented that the organization was wasting more than $250 million a year of donor money, because management could not make the decisions needed to implement what I was recommending. Scrum on paper did not fix an organization that could not decide.

That, to me, is the real story behind “zero payoff.” Rampant incompetence and indecision are going to be cleaned out by AI. The tsunami is coming.

The crippling pattern: AI bolted onto unchanged work

Most companies take AI and attach it to the side of work that hasn’t changed. Someone gets a chatbot in another browser tab. A pilot runs for a quarter. Usage dashboards fill up with logins and tokens. Then a CFO asks what we got for this, and nobody can answer, because nobody measured delivered value before or after.

Three things are almost always missing:

  1. Measurement of delivered value. Companies count prompts, tokens, and seats instead of finished, accepted work. Tokens are telemetry, not value. If token spend is your main AI metric, Slok is right: what you see is a cost center.
  2. Accountability for delivery. AI work sits outside the team that owns the outcome. Nobody is on the hook for turning it into a working product increment, so nobody does.
  3. Feedback loops. Without a short cycle of trying, inspecting, and adjusting, a pilot can’t learn. It drifts until the budget runs out.

What Scrum already gives you in an AI world

Scrum was built for complex work where you can’t predict the answer in advance. Small teams own the result, you measure what they actually deliver, and you adjust every Sprint. AI doesn’t change that. It makes it urgent.

  • Small, accountable teams. An AI agent works on a real team with a real Product Owner and a Definition of Done, not as a side experiment. Accountability stays with people.
  • Empirical measurement. Accepted story points per Sprint, on a stable reference scale, tell you whether a human-AI team is delivering more value, not whether people used AI more.
  • Definition of Done. AI output counts only when it meets the same bar as human output.
  • Sprint Review. Stakeholders inspect a working increment every Sprint, so you learn in weeks, not quarters, whether AI is producing value or just activity.
  • Inspect and adapt. The Retrospective is where a team learns what an agent handles well and what it can’t. Without that loop, there is no learning.

And Scrum forces decisions. Every Sprint, the Product Owner orders the Product Backlog and the Scrum team commits to a Sprint Goal. An organization that can’t decide shows up immediately.

What we learned putting an agent on our own board

My team put an AI agent on our live Jira board as a teammate, with its own account, a role, working agreements, and our Definition of Done. We documented it publicly in an twelve-part series, failures included. (AI Agent as a Real Teammate)

Over nine weekly Sprints from June to August 2026, measured in Jira, the team’s completed points went from 44 to 76 a Sprint, first Sprint to last. The agent’s completed points rose from 3 to 24. Human-assigned output also ended higher, going from 36 to 47 points. No accountability moved from a person to the agent.

The caveats: this was one team, ours. The gain is first Sprint to last, not a steady weekly rate. The counts are based on assignment, so they measure work handed to the agent, not all AI assistance. The biggest lesson wasn’t about the model. An agent is only as productive as the stories it is handed. Backlog quality decided the outcome.

A starting path: the ScrumAI.org beta

That is why we built Agent Bootup, now in free beta at ScrumAI.org. It packages the configuration we spent nine Sprints discovering:

  1. You point us at one machine inside your environment. No new platform license or procurement cycle.
  2. We load the agents remotely and attach them to your Jira board, where your team already works.
  3. The agents take stories off the same Product Backlog as everyone else. Their output shows up in your velocity data as AI points.
  4. We work the Produt Backlog with your Scrum Master and Product Owner, because well-formed stories make agents productive.

Before anything starts, someone on our team replies personally and we have a short fit call. If your Product Backlog isn’t ready, we’ll tell you what needs to change. (Agent Bootup for Scrum Teams)

To be clear about fit: if the board is decorative, the agents will have nothing to pull. If putting one box in your environment takes six months of committee approval, that tells you something too.

The point isn’t to trust our curve. It’s to produce your own, from your own Jira data, so the AI number you take to your executives is yours.

Moving past zero financial payoff: where to start

If your company is one of the S&P 493, you don’t need a bigger token budget. You need one team, one real backlog, one honest velocity measure, and a Sprint Review where you look at what got done.

Apply for Agent Bootup at ScrumAI.org. Places are limited by how many teams we can support at once. If you’d rather talk first, book a consultation with me.

The companies that come out ahead won’t be the ones that bought the most AI. They’ll be the ones that changed how they work, and made the decisions, so AI could deliver.

— Jeff Sutherland


References

  1. Torsten Slok, “No Signs of AI Payoff Outside the Tech Sector,” The Daily Spark, Apollo, Oct. 24, 2026. https://www.apollo.com/wealth/insights-news/insights/daily-spark/no-signs-of-ai-payoff-outside-the-tech-sector
  2. LinkedIn News, “AI has delivered ‘zero payoff,’ says Apollo’s top economist,” Oct. 25, 2026. https://www.linkedin.com/news/story/ai-has-delivered-zero-payoff-says-apollos-top-economist-9568618/
  3. Anthony Ha, “AI delivers ‘zero financial payoff’ to most companies, Apollo economist says,” TechCrunch, Oct. 25, 2026. https://techcrunch.com/2026/10/25/ai-delivers-zero-financial-payoff-to-most-companies-according-to-apollo-economist/
  4. “AI’s boost to the U.S. economy is nearing its peak, Goldman says, as Apollo finds ‘zero financial payoff’ outside Big Tech,” Fortune, Oct. 26, 2026. https://fortune.com/2026/10/26/ai-zero-financial-payoff-outside-big-tech-apollo-slok-goldman-peak/
  5. “Apollo exec says AI has ‘zero financial payoff’ beyond tech sector,” CFO.com, Oct. 27, 2026. https://www.cfo.com/news/apollo-exec-says-ai-has-zero-financial-payoff-beyond-tech-sector/872795/
  6. Ivan Yotzov, Jose Maria Barrero, Nicholas Bloom, et al., “Firm Data on AI,” NBER Working Paper 34836, 2026. https://www.nber.org/papers/w34836 (full text: https://www.nber.org/system/files/working_papers/w34836/w34836.pdf)
  7. Boston Consulting Group, “AI Leaders Outpace Laggards with Double the Revenue Growth and 40% More Cost Savings” (The Widening AI Value Gap: Build for the Future 2025), Sept. 30, 2025. https://www.bcg.com/press/30september2025-ai-leaders-outpace-laggards-revenue-growth-cost-savings
  8. Lindsey Wilkinson, “AI project failure rates are on the rise: report,” CIO Dive, Mar. 14, 2025 (S&P Global Market Intelligence, Voice of the Enterprise: AI & Machine Learning, Use Cases 2025). https://www.ciodive.com/news/AI-project-fail-data-SPGlobal/742590/ ; S&P report: https://www.spglobal.com/market-intelligence/en/news-insights/research/ai-experiences-rapid-adoption-but-with-mixed-outcomes-highlights-from-vote-ai-machine-learning
  9. MIT NANDA, The GenAI Divide: State of AI in Business 2025. https://mlq.ai/media/quarterly_decks/v0.1_State_of_AI_in_Business_2025_Report.pdf ; coverage: Fortune, Aug. 18, 2025. https://fortune.com/2025/08/18/mit-report-95-percent-generative-ai-pilots-at-companies-failing-cfo/
  10. ScrumAI.org, “AI Agent as a Real Teammate: The Complete Series.” https://jvsmanagement.com/ai-agent-real-teammate-complete-series/
  11. ScrumAI.org, “AI Will Not Take Your Job. Someone Who Runs the Agents Will” (Agent Bootup for Scrum teams). https://jvsmanagement.com/ai-agents-for-scrum-teams/
  12. ScrumAI.org, “Your Executives Want an AI Number. Here Is One You Can Defend.” (Agent Bootup for engineering managers). https://jvsmanagement.com/ai-agents-for-engineering-managers/