AI-linked layoffs are no longer a Silicon Valley story — and the count keeps growing. A Reuters factbox tracking AI-linked job cuts, updated 22 October 2026, tallies announcements made since October 2025 across tech, finance and consumer services. At the top sits HSBC, which said in March 2026 it would eliminate about 20,000 roles — roughly 10% of its workforce — as an AI overhaul begins. Amazon follows with 16,000 corporate job cuts announced in January 2026, described as an AI- and efficiency-driven restructuring. Barclays outlined about 5,000 reductions focused on support functions, aiming to save more than £1.1 billion. American Express announced about 3,500 global job cuts on 24 October 2026, a 5% workforce reduction tied to restructuring and AI investment, with a $336 million pre-tax charge. Some reports compute about 3,840 from 5% of AmEx's roughly 76,800 staff; this article uses the Reuters-carried count of about 3,500 and notes the gap rather than averaging two figures.
| At a glance | |
|---|---|
| Source | Reuters factbox of AI-linked layoffs, updated 22 October 2026 |
| Period covered | Announcements since October 2025 |
| Largest counts | HSBC about 20,000; Amazon 16,000; Barclays about 5,000; AmEx about 3,500 |
| Other notable counts | Microsoft 4,800 (July); Cisco fewer than 4,000 (May); Block more than 4,000; HP 4,000–6,000 by 2028; Intuit about 3,000; Standard Chartered more than 7,000 over four years |
| Roles most exposed | Administrative, operational, back- and middle-office, routine coordination and processing |
| Counter-trend | Demand rising for specialised AI engineering and data-science talent |
## Why it matters
The factbox matters because it shows the same playbook spreading beyond tech. Banks feature prominently: Standard Chartered outlined more than 7,000 cuts over four years in May 2026, Mizuho plans up to 5,000 reductions over a decade, Sweden's SEB flagged up to 2,100 cuts by end-2027, Danske Bank about 420 and DNB about 400. Beyond finance, British American Tobacco said in June 2026 it would cut 5,500 jobs and move 3,500 roles to third parties, Dow announced 4,500 cuts (13% of its workforce), and Nike 775. Microsoft announced 4,800 cuts in July, about 2.1% of its workforce, affecting commercial and Xbox businesses. Cisco said in May it would cut fewer than 4,000 roles and expects pre-tax charges of up to $1 billion. Block said it would cut more than 4,000 jobs, nearly half its workforce. The pattern is consistent: firms describe AI overhauls as reorganisations, then use them to justify cuts, shifting budgets from payroll toward data centres, chips and new systems. Meta is the clearest version of the trade-off, flagging that headcount could shrink by up to 20% while it plans to spend $600 billion on data centres by 2028.
## What this is — and what it is not
Analysts covering the factbox stress that "AI job losses" usually blend several things: automation, cost-cutting, and broader reorganisations that move spending from payroll toward data, cloud infrastructure and model development. That distinction matters for jobseekers. This does not look like a single moment where machines replace workers. It looks more like a cost-base reset: fewer people doing routine coordination and processing, slimmer teams with higher output targets, and pressure on management to prove savings quickly. At the same time, the same reports note AI creating demand for specialised engineering and data-science talent. Traditional roles are shrinking while AI roles grow — which is why the useful question for a worker is not "will AI take my job" but "which parts of my job are standardisable, and what adjacent skill moves me toward the hiring side."
## Which jobs are hiring instead
The same companies cutting administrative roles are hiring AI talent at the same time — that is the other half of the factbox. Demand is rising for specialised AI engineering and data-science roles: machine-learning engineers, data scientists, AI infrastructure engineers, and domain specialists who can put AI tooling to work in finance, healthcare and operations. The hiring bar is specific, not generic: employers want people who can build, deploy and measure AI systems, not people who have merely used an AI chatbot. For a worker in an exposed role, the credible move is a concrete adjacent skill — data analysis, automation tooling, or AI-assisted workflows inside your current domain — rather than a vague promise to "learn AI". Short, project-based proof of that skill beats another certificate: a small automation you shipped at work, a data project with real numbers, or a workflow you redesigned.
## What to do next
First, audit your own role honestly: if most of your week is routine coordination, processing, or back-office work that follows a fixed template, that work is the most exposed — not because a model does your whole job, but because a smaller team plus software can absorb it. Second, move toward the hiring side deliberately: AI engineering, data science, and roles that combine domain knowledge with AI tooling are where budgets are flowing. Third, if you are laid off, treat the separation terms and the next application as separate tasks: confirm severance, notice and references in writing, then apply while the market still distinguishes between "cut in an AI restructuring" (common, understood by recruiters) and a performance exit. Fourth, never pay for a "guaranteed AI job" course or placement: the hiring that exists runs through normal applications, and paid guarantees are a scam pattern. The Reuters tally will keep updating — this article's figures are dated to the 22 October 2026 factbox, and readers should check for newer editions before quoting counts.