The headline version of the 2025–2026 layoff wave is simple: AI is replacing workers. Companies are automating, jobs are disappearing, and this is the beginning of a structural transformation in the labor market.

The actual version is more complicated — and more strategically useful.

Of the 1.2 million employees in the US who lost their jobs in 2025, only 55,000 layoffs were attributed to AI, according to outplacement firm Challenger, Gray & Christmas. That’s 4.5% of the total. JP Morgan Asset Management’s analysis of the same data found that the dominant drivers were traditional factors: cost-cutting, slower demand, restructuring, and post-pandemic normalization. Oxford Economics noted that companies sometimes use AI as a “PR explanation” for layoffs that are actually caused by overhiring or weak demand — because citing AI sounds strategic to investors rather than admitting business underperformance.

This matters for strategy because “AI is eliminating jobs” and “the post-pandemic overhire correction is continuing while capital is being reallocated to AI infrastructure” have very different implications for how you manage your business, your workforce, and your competitive positioning over the next three to five years.

The Three Actual Drivers

The post-pandemic normalization. Many tech companies significantly expanded headcount in 2020 through 2022, when demand was elevated and capital was cheap. The 2023 correction was the first wave of reversal. The 2025–2026 layoffs are, in significant part, the continuation of that correction — particularly in companies that delayed the reckoning because they had sufficient cash reserves to absorb the excess capacity longer.

At least 127,000 workers at US-based tech companies were laid off in 2025. The pattern is consistent: companies that grew aggressively during the low-interest-rate period are right-sizing their cost bases to match revenue realities under higher capital costs. The announcement often cites AI or strategic transformation. The underlying driver is often simpler.

Capital reallocation to AI infrastructure. The companies doing the most aggressive cutting are simultaneously making the largest investments in history. As of May 2026, Meta, Amazon, Microsoft, and Alphabet have collectively committed roughly $725 billion in capital expenditure — a 75% increase over 2025 — almost entirely earmarked for AI data centers, chips, and infrastructure. Oracle laid off up to 30,000 employees immediately after reporting strong quarterly earnings, then allocated billions to AI data center expansion.

This is a structural reallocation: payroll budgets being converted into infrastructure budgets. The human capital is being exchanged for compute capital. The roles being eliminated are in functions where AI tools are delivering measurable productivity improvements — customer service, basic coding, data processing, back-office operations. The roles being created are in AI engineering, infrastructure architecture, and roles requiring complex human judgment that AI can’t yet adequately replicate.

AI-induced productivity gains driving genuine workforce restructuring. This is the driver that the AI narrative is mostly right about, but that’s affecting a smaller portion of the workforce than the headlines suggest. SHRM research found that 15% of US employment is at least 50% automated right now, and 12.6% of roles are at high or very high risk of displacement due to AI-powered tools. Forrester projects approximately 6% of jobs will ultimately be lost to AI and automation by 2030 — about 10.4 million jobs in a 160 million job economy.

This is significant but not apocalyptic — comparable in scale to the Great Recession’s job losses, distributed across a longer time horizon and likely offset in part by job creation in adjacent areas. The companies experiencing genuine AI-driven workforce restructuring are the ones that have reached the threshold where AI tools perform specific functions more reliably and cheaply than humans do — customer support at Salesforce and Klarna, coding tasks at Atlassian, logistics coordination at UPS.

“AI Washing” and Its Strategic Implications

The term “AI washing” gained significant traction in 2025 and 2026. It describes the practice of framing layoffs, restructuring decisions, or product pivots as AI-driven when the actual drivers are something else — cost pressure, demand weakness, strategic repositioning.

AI washing has a specific appeal: it reframes organizational failure as strategic transformation. “We overhired and need to cut costs” is a difficult narrative. “We’re realigning to an AI-first operating model” is a compelling one. Investors respond differently to the two framings. Talent responds differently. Media coverage differs.

Approximately 92% of companies that announced AI-driven layoffs actually increased their total headcount between 2024 and 2025. The layoffs were reshuffling, not reduction — cutting in some areas while hiring in others, often ending the period with more employees than they started with but different employees in different roles.

For the leaders of other businesses, the strategic implication of AI washing is specific: when a competitor announces “AI-driven restructuring,” the most useful analysis is not “what AI tools are they deploying?” but “what did they actually need to fix in their business, and how much of this is strategic versus operational?” The answer to that question tells you whether the competitor has genuinely strengthened their position or simply managed the narrative around a correction they were forced to make regardless.

Reading the Wave: A Decision Framework

The current layoff wave is not one thing. It’s three distinct phenomena unfolding simultaneously, with different implications depending on where your business sits relative to each. A framework that treats all of them as “the AI disruption” produces the wrong strategic response to at least two of the three.

Here’s how to position your thinking across the three drivers:

If your primary exposure is driver one — post-pandemic normalization: The strategic question is whether you’ve completed your own correction or whether you’re still carrying cost structure from a headcount base that made sense at 2021 demand levels and 2021 capital costs. Companies that completed this correction in 2023 are entering 2026 with structural cost advantages that translate directly into competitive pricing flexibility, investment capacity, and margin. Companies still carrying the legacy cost base are competing with one hand behind their back. The imperative here isn’t AI — it’s a clean-eyed review of your current headcount against your current and projected revenue at realistic capital costs.

If your primary exposure is driver two — capital reallocation: The strategic question is whether you’re investing in AI infrastructure at a pace that maintains competitive capability, or whether the gap between your infrastructure investment and your most aggressive competitors’ is widening. This isn’t about matching the $725 billion that the hyperscalers are spending. It’s about understanding what AI-enabled capability your competitors will have in 18 to 24 months that you currently don’t, and making a deliberate bet on where to close that gap and where to accept it. Not every AI infrastructure investment by large incumbents translates into competitive advantage for their specific market. Some of it is competitive signaling. The strategic discipline is distinguishing which capabilities will matter for your business from which are irrelevant to your competitive context.

If your primary exposure is driver three — genuine AI productivity displacement: The strategic question is which roles and functions in your business have crossed the threshold where AI tools perform the work more reliably and cheaply than humans, and which haven’t. This threshold varies significantly by function, by industry, and by the specific task structure within a function. Customer support roles with high volume and relatively scripted interactions crossed this threshold earlier than customer support roles requiring significant contextual judgment. Basic coding tasks crossed it before senior architecture decisions. The error that expensive organizations make is applying the aggregate displacement narrative to roles and functions where AI tools haven’t actually reached performance parity — restructuring before the tools can deliver what the restructuring was supposed to enable.

The Klarna case is instructive here. The company was widely cited as a pioneer of AI-first customer service, having dramatically reduced human support staff in 2023 and 2024. By 2025, Klarna was publicly acknowledging that over-reliance on automation had degraded service quality and required reinvestment in human support. The lesson isn’t that AI doesn’t work in customer service — it does, in bounded contexts with well-defined tasks. The lesson is that the threshold for human replacement is specific to the task, not to the function. Applying the displacement decision at the function level, rather than the task level, produces worse service and then a painful and expensive correction.

The Workforce Strategy Implications

Across all three drivers, the current moment has a consistent implication for how thoughtful organizations should be managing their workforce strategy: the distinction that matters most is between capabilities you’re building in people and capabilities you’re substituting with technology.

This is not a binary decision. The organizations navigating this period most effectively are treating it as a portfolio question: for every significant function, explicitly categorizing what work should be done by AI tools, what work should be done by people augmented by AI tools, and what work requires human judgment in ways that current AI tools don’t adequately replicate. The portfolio shifts over time as the tools improve, which means the categories need to be revisited regularly rather than set once and maintained.

The roles that are surviving and thriving share identifiable characteristics: they require complex human judgment that depends on context that is difficult to document, relationship-based work where the value comes from the person rather than the task, creative direction where output quality depends on taste and experience rather than pattern-matching, and strategic planning where the relevant inputs are genuinely uncertain and cannot be adequately specified in advance. Roles with the inverse characteristics — bounded tasks, well-documented processes, high volume, measurable outputs — are the candidates for AI substitution.

The workforce planning question for the next three years isn’t “will AI replace my team?” It’s “which capabilities in my team am I building deeper, and which am I planning to substitute — and on what evidence and timeline?” Organizations that can answer that question specifically, function by function and role by role, are making decisions. Organizations that can’t are reacting to headlines.

The real story of the 2025–2026 layoff wave is not “AI replaces humans.” It’s “capital is being reallocated from labor to compute at an unusual rate, driven partly by genuine AI productivity improvements, partly by post-pandemic normalization, and partly by narrative management.” The businesses that navigate this well are the ones that distinguish between these drivers clearly enough to make deliberate choices — rather than either panic-cutting to appear AI-forward or ignoring the capability shift until a competitor forces the question on them.