This week, Sam Altman publicly said he was “delighted to be wrong” about the AI jobs apocalypse he predicted. The entry-level wipeout he forecast has not materialized. For two years, his timeline anchored boardroom planning. Now, the most authoritative voice on AI labor displacement has openly walked it back.
Two days earlier, Nvidia’s Jensen Huang called the AI-to-job-cuts narrative “lazy” and urged parents to stop chasing AI-proof careers for their kids. The week before that, Standard Chartered’s CEO Bill Winters apologized for framing AI’s purpose at the bank as partly replacing “lower-value human capital.” The cluster is tight: three days, three corrections.
However, you are not Anthropic. You are not Meta. But your owners, your shareholders, or your board have asked you to find ways to reduce overhead, and AI has become the language of that pressure. So you are sitting in the messy middle — maybe you have already cut, maybe you are about to. Meanwhile, the people who built the urgency premise just walked it back publicly.
So the question now is not whether to act. The question is whether the action you are taking maps to a scoreboard that builds enterprise value, or to one that quietly erodes it.
The Wrong Scoreboard in the Technology Lean-out
First, most mid-market CEOs track the wrong things during a technology lean-out.
Specifically, the default scoreboard counts headcount, run rate, burn, and departmental cost. Every CFO produces these numbers. Every board asks about them. They are easy to defend in a quarterly review.
In contrast, the right scoreboard counts capability concentration. Specifically: where judgment actually lives in your organization, not where titles say it lives. Which workflows you have redesigned for human-plus-AI collaboration, and which still run legacy work in legacy form. What augmented productivity baseline you are observing right now — not the theoretical one your vendor promised, but the one you can document with the people you have.
In other words, the first scoreboard tells you what you spent last quarter. The second tells you what you will be able to do in eighteen months.
What Boards Actually Ask For
Critically, a May 2026 IBM Institute for Business Value study found that 98% of CIOs report intensified board pressure to demonstrate AI return on investment since 2024. Meanwhile, 85% say a lack of traceability and explainability has delayed production deployment. In other words: boards want the numbers, operators do not have a clear way to produce them, and the easiest defensible number becomes the cost cut.
That is how the wrong scoreboard becomes the operating plan.
Two Cases of Getting the Technology Lean-out Wrong
In practice, the technology lean-out fails in two predictable ways. Most failures map to one of these patterns.
When IT Drives Alone: McDonald’s and the Missing P&L
In June 2024, McDonald’s quietly ended its three-year, hundred-restaurant test of IBM-powered drive-thru voice ordering. The company killed the technology not because it failed to work, but because no one had ever built the business case for why it should work in the first place.
Specifically, IT led the implementation. The metrics were technical: order accuracy, response latency, integration uptime. The broader operating questions — does this actually reduce labor cost meaningfully? Does it move same-store sales? Does it improve throughput at the scale that matters? — the team tracked loosely or skipped.
When the rollout produced viral customer complaints, the technical defenses did not matter. The P&L story was not strong enough to defend the project. So McDonald’s ended it.
This is what happens when IT drives the implementation alone. The company invests significant time and capital. The technology more or less works. The shareholders never see a clear return.
When the Boardroom Drives Alone: Klarna and the Missing Process
In December 2023, Klarna’s CEO Sebastian Siemiatkowski announced a sweeping shift: the company would freeze all non-engineering hiring and let AI absorb the work. By 2024, an OpenAI-powered customer service agent was handling the equivalent volume of 700 employees. Klarna cut its workforce by roughly 40%. The early numbers looked extraordinary — ten million dollars in claimed annual savings, customer service requests handled at scale, a public case study touted at AI conferences.
In May 2025, Siemiatkowski reversed course. “We focused too much on efficiency and cost. The result was lower quality, and that’s not sustainable,” he told Bloomberg. Klarna is now rehiring humans in customer service, piloting an Uber-style freelance structure to replace the few thousand outsourced agents the company had been running on.
However, the executive strategy was visionary. The execution was un-designed. No one had built the operating process for human-AI handoff. No one had measured what AI could not handle — the empathy, the nuanced problem-solving, the brand-defining moments — until customer satisfaction was already eroding. The savings appeared on the income statement immediately. The damage to retention and brand showed up later, on a different line.
This is what happens when the boardroom drives the implementation alone: the strategy is sharp, the announcement is bold, but the actual operating model does not exist below the slide deck.
What the Scoreboard Should Measure
So the messy-middle CEO needs a different set of questions before any cut. Three of them.
One. Where in this function does judgment actually live — not by job title, but by actual decision-making? If you cut a level on the org chart, can you still answer that question by name? Most CEOs cannot, because the title structure does not match the decision structure.
Two. Which workflows have you redesigned for human-plus-AI collaboration, and which still run as legacy work in legacy form? A function running legacy workflows with new tools is not lean. It just runs short on people.
Three. What augmented productivity baseline are you actually observing right now? Not what the vendor said. Not what your pilot deck claimed. What is the function actually producing now, with the people you have, doing the work the way the new operating model requires?
If you cannot answer those three questions before you cut, you are not cutting from strength. You are cutting from cost pressure on a stale premise — the same premise Sam Altman just walked back this week.
Where the Technology Lean-out Hits the P&L
Specifically, judgment, institutional knowledge, and cross-functional context are off-balance-sheet assets. When you cut them, the loss does not show up the way a layoff shows up — as a one-time charge that flows through the income statement and clears. Instead, it surfaces as a non-cash write-down twelve to eighteen months later: missed opportunities, slower deal cycles, weaker negotiations, customer churn, eroded brand reputation. The board does not see it because it never appears on the scoreboard.
Furthermore, a 2026 report from HRD Connect found that new burnout patterns are not workload-driven. Cognitive overload, fragmented digital ecosystems, constant context-switching, and the emotional strain of adapting to rapid technological change now drive them. Supportive line management can cut burnout by more than half. The capability you fire today becomes the burnout your stretched team faces in eighteen months.
How SiriusXM Got It Right
By contrast, SiriusXM offers the counter-example. The company has spent two years on what its Chief People & Administrative Officer Faye Tylee calls a skills-based transformation. Specifically, the work reshapes thousands of roles around skills, not titles. “Far from complete, and may never be,” she told the HR Brew Talent Summit. The company did not cut first and figure it out later. SiriusXM redesigned first and let the structure follow.
That is what running the right scoreboard looks like in practice.
What This Means for Real Leaders in Plain English
You are feeling pressure to do something. Your owners or your board want a number. AI has become the language of that number. You know that cuts will satisfy the pressure for a quarter or two. You also know — quietly — that the people you would cut hold context the organization cannot easily rebuild. That tension is not a personal failure. It is a structural one.
Specifically, your board is asking you to make a balance sheet decision with an income statement instrument. The right instrument does not exist yet inside most mid-market companies, because no one has built the muscle to see capability concentration as an asset class. Your CFO has no training in it. Your COO has no training in it. Your IT team has no training in it. Your board has not yet learned to ask for it.
That is the conversation worth having. Not where do we cut. Not how fast do we move. But: what do we actually have here? What does the redesigned operating model need to look like? What is the cleanest path from where we are to where we are going? That conversation pays for itself.
Where Alchemy Comes In
Notably, the two failure patterns above share a single root cause. Nobody inside the organization has the external operating perspective to see both the IT side and the executive side of the implementation at the same time. IT sees workflow. The board sees strategy. Almost no one sees how those two layers connect to the P&L. As a result, cuts that ignore the connection produce the very losses they aim to prevent.
Specifically, Alchemy partners with mid-market and founder-led companies at this exact inflection point. We bring an operator’s read of how judgment actually flows in your organization. We also bring the financial frame your board needs to defend the path. We work alongside exclusive development partners. As a result, the organizational redesign and the technology build move in parallel as a single integrated transformation — not as sequential handoffs that lose value at every transition.
If You’re Ready for a Conversation
If you are sitting in the messy middle of a technology lean-out, the conversation starts with the scoreboard. Specifically: are you running the one that builds enterprise value, or the one that quietly erases it?
Book a free discovery call here
References
- TechRadar (May 2026). “Altman: ‘I’m delighted to be wrong.'” https://www.techradar.com/pro/im-delighted-to-be-wrong-sam-altman-says-ai-wont-lead-to-a-jobs-apocalypse
- Business Insider (May 2026). Jensen Huang on AI education advice and the “lazy” job-cuts narrative. https://www.businessinsider.com/nvidia-jensen-huang-what-kids-should-study-ai-education-advice-2026-5
- IBM Newsroom (May 4, 2026). “IBM Study: CEOs are Reshaping C-Suite Roles for the AI Era.” https://newsroom.ibm.com/2026-05-04-ibm-study-ceos-are-reshaping-c-suite-roles-for-the-ai-era
- CNBC (June 17, 2024). “McDonald’s to end AI drive-thru test with IBM.” https://www.cnbc.com/2024/06/17/mcdonalds-to-end-ibm-ai-drive-thru-test.html
- Fortune (May 9, 2025). “Klarna plans to hire humans again, as new landmark survey reveals most AI projects fail to deliver.” https://fortune.com/2025/05/09/klarna-ai-humans-return-on-investment/
- HRD Connect (Dec 2025). “Burnout is back: how organisations can reset for a healthier, more sustainable 2026.” https://www.hrdconnect.com/2025/12/04/burnout-is-back-how-organisations-can-reset-for-a-healthier-more-sustainable-2026/
- Deloitte Insights (2026). “Decision making with AI: 2026 Global Human Capital Trends.” https://www.deloitte.com/us/en/insights/topics/talent/human-capital-trends/2026/decision-making-with-ai.html
- Gallup (2026). State of the Global Workplace. https://www.gallup.com/workplace/349484/state-of-the-global-workplace.aspx
- WSJ via HR Brew (May 2026). Standard Chartered CEO Bill Winters apologizes for “lower-value human capital” remark. https://links.morningbrew.com/c/P-e
Estimated reading time: 9 minutes