Bank Layoffs 2026

InclusionA programme enters when the bank, or named reporting quoting it, gives a role count and a completion date in or through 2026. Unconfirmed figures, redeployment plans and contractor-only cuts sit on the watchlist.

ReadingThe largest number is the oldest programme. Citigroup's 20,000 was set in January 2024 and completes at the end of 2026; the announcements from NAB, Barclays and American Express added 11,000 in four days, October 20 to 23, 2026.

SourceBarclays, Standard Chartered, ABN AMRO, NAB, ANZ and American Express documents; Citigroup filings and call; Reuters, Banking Dive, Banken.nl and the Finance Sector Union for counts not in a company document.

FormulaGap = Tier 3 ÷ (Tier 1 + Tier 2) = 29,800 ÷ 23,290 = 1.28. Same formula as the Axis AI Layoff Tracker, where the all-industry reading is 0.604. Banks run at 2.12× that ratio.

ReadingThe gap is a disclosure choice, not a technology gap. Naming AI as the cause of job losses invites supervisory questions on operational resilience, model risk and conduct. “Simplification” invites none, so the AI link lives in slides and speeches, never in the filing.

SourceTiers assigned by Axis Intelligence Research on each bank's own documents and named executives, not on third-party commentary.

FormulaBACV = (roles ÷ workforce × 100) ÷ programme years. Calculated only where a completion horizon is disclosed, so American Express and Commonwealth Bank carry no reading.

ReadingTwo playbooks. Banks citing AI spread cuts over three to four years through attrition and non-replacement. Banks avoiding the word book one-year restructurings with charges up front. The second group sets 2026 headcount; the first sets 2028.

SourceAxis Intelligence Research, from the role counts, workforce bases and horizons in the ledger. Tier 2 median 3.25 (mean 3.80); Tier 3 median 5.97 (mean 5.63).

Formula19,000 × 0.15 = 2,850 FTE-equivalents against 5,000 × 0.60 = 3,000 support-function roles. A comparison of scale, not a claim that one caused the other.

ReadingThe productivity Barclays reported in its 2025 annual report was already large enough to absorb this cut without lost output. Barclays now says so itself: fewer developers “as AI makes our people more productive”.

SourceBarclays Form 20-F 2025 (developers, productivity, c.100,000 Copilot licences); Q3 2026 results announcement, slides and management speech; H1 / December split from the analyst call as reported by Banking Dive.

Formula(205,198 − 194,369) ÷ 23,762 = 45.6%. Net headcount against announced cuts: two different measures, set side by side to show the scale of what announcement counts do not capture.

ReadingThe quarterly decline is slowing, from −5,623 to −3,097, but has not stopped. In September the CFO said AI “will bring headcount down more”. In the US, the earnings supplement is the better layoff tracker.

SourceWells Fargo Q3 2026 Form 8-K, Exhibits 99.1 and 99.2; Q3 2026 earnings call; Challenger, Gray & Christmas, September 2026 report. Ratios by Axis Intelligence Research.

FormulaRegional share = regional roles ÷ ledger total. Scenario: 53,090 + 20,000 = 73,090, with HSBC added to the UK. The toggle never changes the confirmed ledger.

ReadingThe UK is the only region where every confirmed role carries an AI attribution. If HSBC confirms, the UK overtakes the United States and becomes 44.9% of the total.

SourceAxis Intelligence Research. HSBC figure from Bloomberg reporting of March 19, 2026, citing unnamed sources; not confirmed by HSBC.

ScaleMarks count the distinct banks in the dataset with a documented action on that function, out of three.

ReadingRelationship banking is almost untouched. The cut falls on work that was already standardised, centralised or moved offshore, which is exactly the work that automates first.

SourceCompany documents and named reporting listed per bank in the dataset.