When Intel told several thousand engineers in its foundry division this April that their positions were being eliminated, the company pointed to the same rationale Microsoft had used in January when it cut roughly 1,000 workers, and the same one Meta had invoked in March when it shed 1,200 roles from Reality Labs and recruiting: the need to redirect resources toward artificial intelligence.
The language varies slightly from one earnings call to the next. “Reallocating toward AI priorities” is a favorite. So is “investing in the AI transition.” But the pattern is consistent, and it is accelerating. According to Layoffs.fyi, the crowdsourced tracker maintained by Roger Lee that has become a go-to reference for journalists and labor researchers, at least 128,270 tech workers have lost their jobs so far in 2026. That count, current as of late May, spans publicly traded giants, mid-sized software companies, and venture-backed startups. It almost certainly undercounts quiet reductions that never reach the press.
The same companies driving those cuts are simultaneously making the largest infrastructure bets in the industry’s history. Meta’s 2026 capital expenditure guidance sits between $60 billion and $65 billion, nearly all of it earmarked for AI data centers and custom silicon. Microsoft has committed roughly $80 billion to AI-enabled data centers in its current fiscal year. Alphabet disclosed approximately $75 billion in 2025 capital expenditure and has signaled that pace will hold or increase through 2026. Amazon’s AWS infrastructure spending runs on a comparable track. Factor in Apple, Oracle, and a tier of heavily funded AI startups, and analyst projections from firms including IDC and Goldman Sachs place total industry AI-related capital spending in the range of $700 billion to $750 billion for the year. The $725 billion figure in the headline reflects the midpoint of those estimates, not a single audited total.
Where the Cuts Are Landing
The layoffs are not spread evenly across the workforce. Operations, customer support, quality assurance, and mid-level software engineering roles have absorbed a disproportionate share of the reductions, based on company announcements compiled by Layoffs.fyi and reporting from outlets including The Wall Street Journal. Teams focused on AI research, machine learning infrastructure, and data-center operations have largely been spared or actively expanded.
That pattern holds regardless of company size. At large platforms, restructuring announcements pair headcount reductions with pledges to double down on AI. At startups, the dynamic is blunter: founders who cannot demonstrate an AI-native product thesis are struggling to raise follow-on rounds, and the resulting cash crunches translate directly into layoffs.
Roger Lee, whose work building Layoffs.fyi was profiled by The New York Times, started the tracker during the pandemic-era downturn. It gained wider attention through the massive 2022 and 2023 waves, when Amazon, Google, and Meta each cut more than 10,000 workers. By Lee’s count, the 2026 cycle is on pace to rival those years in total volume.
A $725 Billion Bet With Fuzzy Receipts
There is a significant gap between what companies say about AI-driven efficiency and what the available data can actually prove. Earnings calls and shareholder letters routinely pair layoff announcements with updates on AI infrastructure investments, creating an implied causal link: we are cutting people because AI can do the work. But company filings typically group restructuring charges together without isolating the portion tied to automation. Layoffs.fyi does not specify whether any given job was eliminated because of AI deployment, a revenue shortfall, or a broader cost reduction.
No government agency has published a standalone breakdown of 2026 tech layoffs by role, company, or cause. The Bureau of Labor Statistics tracks mass layoff events and unemployment claims, but its industry categories do not map neatly onto “tech” as the sector defines itself. That leaves researchers, policymakers, and displaced workers piecing together a picture from partial sources: a crowdsourced tracker, selective analyst notes, and company press releases crafted to reassure investors.
The $725 billion spending figure carries its own uncertainty. Analyst projections from IDC, Gartner, and Wall Street banks rely on proprietary models and varying definitions of what qualifies as “AI spending.” A dollar spent on a data center that runs both traditional cloud workloads and AI training jobs could be classified either way depending on who is counting. The number is directionally credible, given the scale of disclosed capex from the five largest spenders alone, but it remains a projection.
What Displaced Workers Are Up Against
For the people behind the 128,270 figure, the ambiguity around causes matters less than the reality of a tightening job market. Job boards in software engineering, product management, and IT operations are more crowded than at any point since early 2023, according to hiring data tracked by platforms including LinkedIn and Indeed. Severance packages at major firms have generally held at 12 to 16 weeks, but re-employment timelines have stretched considerably.
On Blind, the anonymous workplace forum popular among tech employees, and in layoff-specific communities on Reddit, mid-career engineers report fielding fewer recruiter messages and facing longer interview cycles, particularly for roles that do not involve AI or machine learning. The contrast is stark: companies are hiring aggressively for GPU cluster management, large language model fine-tuning, and AI safety research while pulling back on the kinds of positions that employed the bulk of the industry’s workforce for the past decade.
The retraining question looms over all of it. Companies and policymakers have urged displaced workers to “upskill” into AI-adjacent roles, but the pipeline from, say, a customer support management position to a machine learning engineering job is neither short nor cheap. Community colleges and coding bootcamps have launched AI-focused curricula, yet there is little reliable data on placement rates for career-switchers compared to new graduates who entered the field with AI training already in hand.
Organized labor responses remain limited. The Communications Workers of America and the Alphabet Workers Union have called for greater transparency around AI-related displacement, but union density in the technology sector is still vanishingly low compared to manufacturing or public-sector employment. Legislative proposals at the state level, including California’s proposed AI accountability bills, have yet to produce binding requirements for companies to disclose how automation factors into layoff decisions.
Routine Correction or Something Else Entirely
The tech industry has a well-documented habit of over-hiring during booms and cutting aggressively when growth slows. The 2022-2023 contraction followed a pandemic-era hiring spree that added hundreds of thousands of roles across the sector. Some portion of the current layoffs almost certainly reflects that familiar correction rather than a permanent replacement of human labor by machines.
But several factors distinguish this cycle. The sheer scale of capital being redirected toward AI infrastructure has no precedent in the industry’s history. The companies making the deepest cuts are simultaneously reporting strong or growing revenue, which undercuts the argument that layoffs are purely a response to financial pressure. And generative AI tools have reached a level of capability that makes certain categories of knowledge work, from code generation to content summarization to front-line customer interaction, genuinely automatable in ways that were theoretical two years ago.
If AI products fail to deliver the productivity gains their backers promise, companies may find themselves short-staffed and scrambling to rehire. The history of technology adoption is full of premature declarations that human workers are obsolete. But the combination of record investment, accelerating layoffs, and explicit corporate messaging tying the two together makes 2026 feel less like a routine correction and more like the opening phase of something longer and less predictable.
For now, 128,270 is a number on a tracker. Behind it are engineers, designers, project managers, and support staff whose careers have been interrupted by a bet their former employers are making on a technology whose returns are still unproven. The capital markets are confident. The workers those markets displaced are still waiting for evidence.
