Commentary
Last week, Microsoft and Meta both announced significant reductions in force. Big Tech layoffs—sometimes tens of thousands of employees at once—make for attention-grabbing news. They portend imminent doom: human displacement driven by artificial intelligence (AI).
But behind the frightening headlines, Big Tech is growing revenue and expanding profit margins, all while increasing total headcount and investing astronomical sums in capital expenditures geared toward AI and automation.
Since 2022, technology companies have announced roughly 750,000 layoffs in the United States. The peak came in 2023, when 264,000 jobs were eliminated, and 2025 nearly matched that peak. Another 96,000 layoffs have been announced in the first four months of 2026. By conventional measures, this would imply the sector’s in sustained retrenchment.
Except Big Tech isn’t cutting back. It is growing. Net employment—the net number employed in the sector, accounting for both cuts and hires—has grown every year through this period. In 2022, total employment grew by 286,000. In 2023, the year of maximum headline carnage, the sector still added 117,000 jobs net of layoffs. In 2024, net tech employment grew by 300,000. The sector ended 2024 with approximately 9.9 million workers—a historic high.
During the pandemic years of 2020 and 2021, Big Tech undertook a hiring binge, roughly doubling headcount. Companies like Meta, Amazon, and Alphabet grew their workforces at rates with little connection to sustainable business models, more akin to speculative expansion. This was inevitably going to reverse. The 2022–2023 layoff wave was not AI displacement. It was a hangover.
What followed—and what is still unfolding—is a different and more consequential story.
The five largest U.S. technology hyperscalers—Alphabet, Amazon, Apple, Meta, and Microsoft—generated a combined $1.55 trillion in revenue in 2022 with an aggregate operating margin of roughly 19 percent. By 2025, revenue had grown to $2.1 trillion and operating margins had expanded to 27 percent. These companies were simultaneously cutting bloated administrative headcount, growing revenue, and expanding profitability. This is not a contradiction. It is an efficiency thesis playing out in the numbers: fewer high-cost, low-productivity employees, more output, better margins.
The operating cost compression was deliberate and investor-driven. After years of rewarding growth at any cost, capital markets pivoted sharply around 2022 toward margin discipline. Executives responded accordingly. Middle-management layers were flattened. Coder-to-non-coder ratios were tightened. Corporate support functions were trimmed. Investors cheered. The result was a margin recovery that made the layoff wave look, in retrospect, like business optimization rather than crisis.
But the more important number is not headcount at all. It is capital expenditure (CapEx)—the investment in physical infrastructure and equipment that companies make for their future.
In 2022, the CapEx of those same five hyperscalers totaled a combined $162 billion. By 2025, that figure had reached $448 billion—a near-tripling in three years. CapEx as a share of revenue, which had been running at 10 percent, hit 21 percent in 2025—the highest in over a decade for an industry that built its valuation premium on being asset-light.
The 2026 guidance numbers are staggering: collectively, these companies have signaled more than $600 billion in CapEx this year, with Amazon alone projecting $200 billion and Alphabet guiding up to $185 billion. Rather than relying on internal cash flow, a substantial portion of this investment is now being financed by debt. Alphabet issued $25 billion in bonds in late 2025; its long-term debt quadrupled in a single year. Oracle already looks over-indebted. Amazon has said it may tap equity and debt markets as the buildout continues. The asset-light era is over.
Nearly all of this spending is directed at AI infrastructure and automation. The hyperscalers are building the foundation for a technology shift they believe will be as consequential as the internet—and they are betting their balance sheets on it. The spending is happening whether or not that bet is correct. If they are wrong, the consequences will be severe.
This brings us to the AI-displacement question that dominates public conversation, and that I addressed last summer. How much of the current Big Tech layoff wave is actually AI eating jobs?
Less than the headlines suggest—for now. According to Challenger, Gray & Christmas, which tracks the reasons companies give for layoffs, AI was cited in approximately 5 percent of all cuts in 2025. Since companies began explicitly attributing layoffs to AI in 2023, the cumulative AI-cited figure is about 3.5 percent of all cuts tracked. So far in 2026, AI ranks fifth among stated layoff reasons by volume, though it reached 25 percent of all cuts in March 2026 alone, suggesting acceleration.
The explicit AI citation number almost certainly understates the real effect. “Restructuring” and “cost reduction” are legally cleaner ways to announce a workforce reorganization driven by AI adoption—companies do not tend to hand plaintiffs’ attorneys a ready-made narrative. Salesforce eliminated more than 4,000 customer support roles in 2025, and its CEO stated publicly that AI was already handling 30 to 50 percent of the work. Amazon’s CEO explicitly warned his workforce that AI would reduce the headcount needed for a range of functions. Microsoft, which has cut approximately 15,000 workers since the start of 2025, framed its most recent round as moving “from a software factory to an intelligence engine.”
We are in the early innings of genuine AI-driven labor reorganization, but the magnitude of displacement so far is modest relative to the scale of the industry—and dwarfed by the post-COVID normalization that preceded it. The story is not yet “AI is eliminating tech jobs.” The story is that AI is being used to do more with the same number of people, which shows up in margin expansion. The roles most clearly threatened are support functions, customer service, and lower-complexity coding tasks, not the engineering and research roles that dominate high-end tech employment.
What is without modern precedent is the combination: the largest sustained capital investment surge in tech history, increasingly funded by debt, occurring simultaneously with the most prolonged layoff cycle since the dot-com bust, while net employment still grows and operating margins expand. These are not contradictory signals from a sector in distress. They are coherent signals from a sector in structural transformation—a sector shedding one model of labor intensity and building the infrastructure for another.
The trendline is not hard to see. Over the next one to two years, AI-attributed layoffs will become a larger share of announced cuts as AI agents move from experiment to deployment in enterprise environments. McKinsey estimates that 32 percent of companies expect a workforce reduction specifically tied to AI within the next year. Gartner projects that 40 percent of enterprise applications will feature AI agents by the end of 2026, against less than 5 percent today. The roles at highest risk are mid-level knowledge workers in support, quality assurance, customer operations, and entry-level coding. The sector will continue to add jobs in aggregate—demand for AI-fluent engineers, data scientists, and cybersecurity specialists is structural and unlikely to slow—but the composition of those jobs will shift toward higher-skilled, higher-compensated roles, and the attrition in middle-skill tech functions will accelerate. This is not a crisis narrative. It is a reallocation narrative.
Still, it behooves all of us to proactively consider how AI will affect our jobs and our futures. There will be winners and losers, both at the corporate level and for individual occupations. It is not a time for fear but for investment in new skills and new ways of working. Those who do so are most likely to come out ahead.
Friends Read Free
Copy
Facebook
Tweet
