Is Your Company Actually Restructuring Around AI?

This is the fourth in a series exploring what the research actually says about AI and your career. Previously: AI displacement is real, which roles are most exposed, and AI is quietly rewriting your role. This week: is your company actually restructuring around AI, or just talking about it?

How to read the signals most professionals miss

Over the past three weeks, we confirmed that AI displacement is real, that which roles are more exposed depends on the makeup of daily work tasks, and that AI is quietly editing roles from the inside by taking specific work tasks first. This week's question is the one people ask when they look up from their own role and scan the room: is my company genuinely restructuring around AI, or just talking about it?

The short answer: probably both. And the gap between the two is where most professionals sit right now.

Most companies adopted AI. Most haven't changed how they operate.

Between 70% and 88% of companies are now using AI in some form, depending on which survey you look at (Barrero et al., 2026; McKinsey, 2025). Your company almost certainly has Copilot licenses, an internal tool, or at minimum teams using ChatGPT informally. AI is in the building.

But if you look around your team and think "not much has structurally changed," you're in the majority. Over 80% of companies report no measurable impact on either employment or productivity from AI over the past three years (Barrero, Bloom, Davis et al., NBER 2026, surveyed Nov 2025 through Jan 2026). A separate study using Danish administrative records found the same pattern: despite 41% of workers in AI-exposed roles actively using ChatGPT, zero measurable effect on earnings or hours worked through 2024 (Humlum & Vestergaard, NBER/PNAS 2025).

Now, this doesn't mean restructuring isn't happening anywhere. It clearly is. The Meta layoffs, Amazon's 16,000 cuts, McKinsey reducing 5,000 roles over five years while deploying internal AI agents: these are real. But those headline cases are concentrated in large technology and information-intensive companies. They represent the roughly 20% of companies where AI has changed operations. For the other 80%, including most mid-size companies, professional services, and non-tech industries, the pattern is different: individual people are saving time on individual work tasks. Someone drafts emails faster. Someone generates reports more quickly. But the team structure stayed the same. Roles weren't eliminated or consolidated. Workflows weren't redesigned. The time savings quietly dissipated into slightly shorter workdays or slightly less stressed weeks.

The technology arrived at most companies. The organizational change, at most companies, didn't follow.

Executives are planning something different from what employees expect

Here's where it gets uncomfortable.

The Barrero et al. study (nearly 6,000 executives across the US, UK, Germany, and Australia, surveyed by the Atlanta Fed, Bank of England, Bundesbank, and Macquarie University) asked both executives and employees what they expect AI to do to employment at their companies over the next three years.

Executives predict AI will reduce employment by 0.7%. Employees at those same companies predict AI will increase employment by 0.5%.

Same companies. Opposite conclusions. The people making restructuring decisions see AI as a way to do the same work with fewer people. The people whose work is being automated see AI as a tool that lets them contribute more, handle bigger projects, deliver more output, and become harder to replace. Both can't be right.

And this isn't just survey sentiment. 32% of companies are planning headcount reductions of 3% or more within the next year (McKinsey State of AI, Nov 2025). Larger companies are more likely to expect enterprise-wide cuts. Meanwhile, middle management job postings have dropped more than 40% between April 2022 and October 2024 (Deloitte, 2025). Companies are compressing management layers, and that compression is already showing up in hiring data even if it hasn't reached your team yet.

Why the gap? Because reorganizing work is harder than deploying tools.

There's a concept researchers call the "productivity J-curve" (Humlum & Vestergaard, 2025). When a major technology arrives, productivity often dips before it rises. Not because the technology doesn't work, but because the organization hasn't changed around it.

Here's a concrete example. AI can draft a quarterly business review document in 10 minutes instead of 2 hours. But if the rest of the process stays the same (the same three rounds of review, the same approval chain, the same distribution list, the same follow-up meeting), the company doesn't capture those savings. You just have a faster first draft sitting in the same slow pipeline. The work task got faster. The way the team operates around that work task didn't.

This is why 80% of companies report no impact. And it connects to something Acemoglu formalized in peer-reviewed research (Economic Policy, 2025): only about 20% of US work tasks are meaningfully exposed to AI, and of those, only about 23% can be profitably automated given current costs, quality requirements, and governance needs. That's roughly 4.6% of all work tasks where the economics justify full deployment today. The gap between "AI can technically do this work task" and "it's worth reorganizing our team around AI doing this work task" is where most companies are stuck.

The companies that broke through (roughly 20%) did something specific. They didn't just give people AI tools. They changed the sequence of who does what, eliminated handoff steps, rebuilt review processes, and reassigned accountability for outputs. In practice, that means things like: the finance team no longer has three analysts compiling data for a fourth analyst to summarize; one analyst works with AI to compile and summarize, while the other three shift to exception analysis and stakeholder communication. The team size may or may not change, but the work each person does is fundamentally different. Companies that made these changes report 20% cost-efficiency improvements (McKinsey State of Organizations, 2026).

And here's a finding that complicates the "restructuring = layoffs" narrative: peer-reviewed evidence shows that the strongest link between AI investment and company performance runs through product innovation, not cost-cutting (Babina et al., Journal of Financial Economics, 2024). The companies benefiting most from AI are growing by creating new products and entering new markets, not primarily by reducing headcount. If your company's AI strategy is purely about efficiency and doing more with less, that's one signal. If it's about building new capabilities and products, that's a different and more positive signal.

How to tell where your company stands

The research points to three phases. Most companies are in the first. The practical question is: which phase is your company in?

Phase 1: Tools deployed, nothing changed (the ~80%)

AI tools are available. Some people use them. No changes to team structure, how work moves through the organization, or headcount. Time savings are real but absorbed informally.

If most answers are yes/no in the expected direction, your company is in Phase 1. The restructuring hasn't started. But with 32% of companies planning 3%+ reductions in the next year, the transition from Phase 1 to Phase 2 can happen faster than it feels from the inside.

Phase 2: How work gets done is changing (~20%)

Leadership has moved beyond tool deployment to changing the way work actually moves through teams. Some roles are shifting. Positions aren't being backfilled when people leave. New AI-related responsibilities are appearing.

If you recognize several of these, the restructuring is underway. Your response should be positioning, not panic: make sure your work is concentrated in the durable, human-essential parts of your role (judgment, relationships, exception handling, domain expertise), not the work tasks being absorbed by AI.

Phase 3: The operating model is changing (early movers)

The company is actively redesigning how it operates around AI. Teams are structured as humans directing AI agents. Management layers are compressed. New roles (AI quality review, agent oversight, AI governance) are being created.

If this sounds like your company, the restructuring is advanced. The opportunities are real (new roles, higher-value work, AI-augmented specialization), but so is the displacement risk for anyone whose work sits entirely in the automatable zone.

The question behind the question

"Is my company restructuring around AI?" is really asking something more personal: how much time do I have, and what should I be doing now?

If your company is in Phase 1, you have a window. Not an indefinite one, but enough to position yourself deliberately. Use the time to shift your work toward the parts of your role that AI can't touch: the judgment calls, the stakeholder relationships, the exception handling, the domain expertise that takes years to build.

If your company is in Phase 2 or 3, the window is smaller. The restructuring is visible and your positioning matters more urgently. The professionals who thrive in these environments are the ones who've already identified their durable strengths and can articulate what they bring that AI doesn't.

Either way, the starting point is the same: knowing which parts of your work are exposed and which are durable.

Research referenced:

1. Barrero, Bloom, Davis et al., "Firm Data on AI," NBER Working Paper 34836, March 2026
2. Humlum & Vestergaard, "Large Language Models, Small Labor Market Effects," NBER WP 33777 / PNAS, September 2025
3. Acemoglu, "The Simple Macroeconomics of AI," Economic Policy, 2025
4. Babina, Fedyk, He & Hodson, "Artificial Intelligence, Firm Growth, and Product Innovation," Journal of Financial Economics, 2024
5. McKinsey, "The State of AI in 2025," November 2025
6. McKinsey, "The State of Organizations 2026," February 2026
7. McKinsey, "The Agentic Organization," September 2025