The 2026 AI Layoff Wave: What Got Cut, What's Coming Next

On 20 May 2026, Meta sent layoff notices to roughly 8,000 employees. The same week, the company reported record quarterly revenue of $56.31 billion and net income of $26.8 billion. The savings will flow into the roughly $145 billion Meta plans to spend on AI infrastructure in 2026. This is the shape of the 2026 AI layoff wave. It is not a company in trouble cutting to survive. It is a company at peak earnings deciding payroll was the wrong place for the money.

The story most newsrooms wrote in 2026 is "AI is replacing your job." The story the data actually tells is more uncomfortable. AI is sometimes replacing jobs. Other times it is the official reason for a cut that would have happened anyway. Often it is both at once. The wave moving through tech, finance, IT services, and media this year is not one phenomenon. It is three, layered on top of each other, sharing the same headline.

This piece is the roundup. It walks through every major confirmed AI-cited layoff in 2026 so far, the actual numbers, what each company said, and what the data shows underneath the press release. By the end you should be able to read the next layoff announcement and tell which of the three things you are looking at.

The aggregate number first, because it anchors the rest. As of 18 May 2026, Layoffs.fyi tracked more than 113,000 tech layoffs across 179 companies this year. Q1 2026 alone saw 81,700 cuts, the highest quarterly total since early 2023. Roughly 20% of those layoffs were explicitly attributed to AI by the company itself. Per Challenger Gray & Christmas, AI is the fifth-most-cited reason for cuts in 2026, behind market conditions, restructuring, closures, and cost reduction. The fact that those four still rank higher matters more than the headlines suggest. (A deeper dive on the underlying statistics is here.)

Every major AI-cited layoff in 2026 so far

The list below is chronological. Each entry gives the company, the number, the official reason, and the source. If a figure could not be verified from a primary source or major outlet, it is qualified or omitted.

Amazon (28 January 2026). 16,000 corporate layoffs. CEO Andy Jassy framed the cut as efficiency from AI: "we will need fewer people doing some of the jobs that are being done today." Cumulative cuts since Jassy's October 2024 memo signaling AI-driven restructuring: more than 30,000 corporate roles. The savings are flowing into AI data-center build-out, where Amazon's 2026 capex is approaching $100 billion.

Salesforce (10 February 2026). Roughly 1,000 layoffs across Agentforce, Heroku, marketing, product management, and data analytics. The deeper number is the support workforce. Over the trailing twelve months, Salesforce reduced customer-support headcount from approximately 9,000 to 5,000 as Agentforce and Einstein Copilot absorbed Tier 1 and Tier 2 work. CEO Marc Benioff stated Salesforce hired no new engineers in fiscal year 2026 because AI coding tools handle the workload. Total headcount, however, sits at a record 83,000. This is reshape, not shrink.

TCS (Tata Consultancy Services), full FY26. Net headcount reduction of 23,460, taking the workforce to 584,519. The net number understates the cuts because TCS continued hiring roughly 40,000 freshers per year. The reduction landed at mid-to-senior levels through "voluntary exits." TCS now uses AI to manage roughly 50% of internal role allocations. AI-related revenue exceeded $2.3 billion in Q4 FY26 alone. For context, the combined top five Indian IT firms (TCS, Infosys, Wipro, HCLTech, Tech Mahindra) had a net headcount decline of 6,981 in FY26, which means everyone except TCS net-grew. TCS is the outlier on the cutting side.

IBM, Q4 2025 through April 2026. The widely-reported "8,000 HR layoffs" figure is wrong. IBM clarified the actual number was in the hundreds, not thousands. Approximately 200 HR roles were eliminated through automation by the AskHR chatbot, which now handles 94% of routine employee queries (payroll, leave, policy). Total US positions eliminated in this window: between 3,000 and 9,000. Cumulative IBM cuts since September 2024: more than 15,000. The dominant driver is not AI but US-to-India cost arbitrage, where median tech salaries run $22,000 against $150,000. IBM is also tripling US entry-level hiring in engineering and client-facing roles per Bloomberg. AI gets the press release. Geographic cost optimization does the work.

Microsoft, fiscal year 2026. Headcount declining year-over-year through Q3 (ending 31 March 2026), per the company's 10-Q filing. Microsoft offered voluntary buyouts to approximately 125,000 employees, the largest single voluntary-departure lever any 2026 major has used. AI and Copilot teams were explicitly exempt from the hiring freeze and from the buyout pool. Cuts have hit content moderation, customer support, software testing, and certain engineering roles. Combined Meta and Microsoft layoffs in April 2026 totaled roughly 20,000.

Meta (20 May 2026). 8,000 layoffs, about 10% of the 78,000 global workforce. Six thousand open requisitions cancelled, raising the effective reduction to roughly 14,000. Seven thousand employees reassigned internally to AI-development roles. Same week: record quarterly revenue of $56.31 billion. Mark Zuckerberg said no further company-wide cuts in 2026, but additional reductions are expected through August and the fall.

Wall Street back office and trading, rolling through 2026. Morgan Stanley cut 2,500 roles across investment banking, trading, wealth management, and investment management. Aggregate headcount across the six biggest US banks (JPMorgan, Bank of America, Morgan Stanley, Wells Fargo, Citi, Goldman) sat at 1.09 million at year-end 2025, the lowest since 2021. A JPMorgan executive told investors that operations and support staff would fall at least 10% over the next five years even as business volume grows, citing AI. AI plus algorithmic trading is replacing entry-level and back-office work first, senior relationship work last.

Business Insider (May 2026). Less than 5% of the newsroom, fourth consecutive year of cuts. The bigger signal is that more than 70% of the remaining newsroom now uses Enterprise ChatGPT for reporting and research, with the company targeting full adoption. The cut is small. The shape is the story.

Klarna, reversed in early 2026 (the counter-example). Worth its own paragraph because it is the strongest evidence the "AI replacement" story is over-claimed. In 2024 Klarna announced that its AI agent had absorbed the workload of approximately 700 customer service representatives, handling 2.3 million conversations in 35 languages. By early 2026 CEO Sebastian Siemiatkowski publicly admitted the move had gone too far. Customer satisfaction had deteriorated on complex interactions. The projected cost savings had not fully materialised. Klarna is now rehiring human agents under a hybrid model where AI handles routine queries and humans handle escalations. Siemiatkowski told Bloomberg that Klarna's AI-focused path of the past few years had not been the right one.

The pattern beneath the announcements

Three patterns are visible across the cuts, and they explain why the "AI is replacing everyone" framing misreads what is actually happening.

Pattern one: AI is cutting roles that are rules-based and transactional. Customer service (Klarna, Salesforce support), HR routine operations (IBM AskHR), bank back-office processing (JPMorgan), basic IT services bench work (TCS mid-level), content moderation (Microsoft), tier-1 support tickets. What these roles share is that the output is fungible. One reply looks like the next. The decision tree is documented. The exceptions are narrow enough to handle by escalation. These roles are getting cut because the work is automatable, not because the workers are replaceable in some deeper sense.

Pattern two: AI is the official reason for cuts that companies wanted to make anyway. This is the harder thing to see, but the data is consistent. Per Challenger Gray & Christmas, AI ranks fifth among the cited reasons for 2026 layoffs, behind market conditions, restructuring, closures, and cost reduction. Companies have wanted to flatten HR, middle management, and back-office functions for at least a decade. AI provided the permission structure to do it now. The IBM US-to-India arbitrage is the cleanest example. The cut is geographic and structural. The press release is about automation.

Pattern three: Companies are converting payroll into capex at peak earnings. This is the 2026 pattern that 2023's tech layoffs did not have. Meta cut 8,000 while reporting record revenue. Microsoft is buying out 125,000 employees while AI-team headcount grows. Salesforce stopped hiring engineers entirely. The cumulative AI infrastructure spend across hyperscalers in 2026 is approximately $725 billion. Some of that is genuinely productive investment. Some of it is a bet, and the bet is being financed by the people whose work the bet might eventually obsolete. Q1 2026 was the worst quarter for tech layoffs since early 2023, even though 2026 tech revenue and profit are at record highs. That gap is the story.

What is not getting cut at scale, and why. Relationship-bearing roles where the customer pays a premium for the named person. Judgment-and-blame-bearing roles where someone has to sign the regulatory disclosure or the clinical chart. Roles that have already been cut to bone in prior restructuring cycles and have nothing left to give. The longer breakdown of roles AI will not replace covers the full list.

What this means for your career

If your role appears in the cut list, three things follow.

The 12-to-18-month window is real. The hiring market for the roles getting cut narrows from both sides. Companies eliminate the role internally, and other companies stop posting it externally. The window to reskill or reposition is now, not when the next round hits. The longer list of which jobs are getting hit first covers the order.

Reskill toward reviewer-not-doer. AI does the underlying task. Humans review, escalate, sign, and sit with the customer when the AI gets it wrong. The roles that survive in the affected functions are the ones where the worker became the reviewer of AI output instead of competing with it. This is the cleanest career move available right now. It does not require a new degree. It requires a deliberate twelve months of using the tools in the work you already do.

Geography matters more than it did. The 2026 cuts are concentrated in cost-pressured markets. US tech, UK financial services, the large Indian IT majors. Smaller and slower markets are running twelve to eighteen months behind on the cutting cycle. That is not permanent insulation. It is a window.

If your role does not appear in the cut list, three other things follow.

Do not assume safety. The wave moves. In 2023 the cuts were entry-level engineering and recruiting. In 2026 they are mid-level support, basic legal review, junior dev, accounts payable, low-tier HR. The function that looks insulated this year is the function that gets restructured next year if the AI capability matures.

Watch the leading indicator. When your function's headcount stops growing while company revenue keeps growing, that is the year-out signal. Look at the gap between revenue growth and headcount growth in your specific function. Salesforce's gap (record total headcount, zero new engineers) is what the start of a wave looks like in the data before it shows up in the press release.

Score it for your role, not the occupation average. A generic "your job is X% at risk" number is exposure, not displacement. What you need is the personalised score that weights the dimensions for your role, industry, country, and seniority.

How RiskQuiz scores this

The RiskQuiz methodology scores nine dimensions across role, industry, country, seniority, function, employer type, automation exposure, and a few others. It uses McKinsey adoption data, the Anthropic Economic Index, BLS occupational projections, and the same Challenger layoff attribution data referenced earlier in this piece. The output is a 0-to-100 risk score with the dimensions weighted for your specific situation, not a generic occupation average. If you took the quiz a year ago, your score has probably moved, because the 2026 wave changed the weightings on several dimensions. The quiz takes 4 minutes. The report explains the score, the dimensions that drove it, and what to do about it for your role.

What to do about it

The 2026 AI layoff wave is real, smaller than the headlines suggest, and uneven. Some roles got cut because the work automated. Some roles got cut because the company wanted them gone and AI was the cover. Some roles got cut while their company posted record profits. The pattern is not the same for everyone. The risk is not the same for everyone.

The quiz takes 4 minutes. The personalised report explains, in your role and industry context, what the next 18 months look like. Take it here.