AI Layoff Tracker 2026: Every Job Cut Companies Have Actually Documented as AI-Driven

AI Layoff Tracker 2026

Last updated: October 4, 2026, 09:00 UTC — Updated weekly (Fridays), and within 48 hours of any qualifying regulatory filing. | Maintained by: Axis Intelligence Research & Davis Park

Why this tracker exists

At least five public trackers currently count “AI layoffs.” They do not agree with each other, and the gap between them is not small — published 2026 totals range from roughly 96,000 to over 316,000 for overlapping periods. The reason is not arithmetic. It is that none of them define what “attributed to AI” means, so each is silently counting a different thing.

This tracker separates job cuts by who made the AI claim and in what document. A statement in an SEC filing carries different evidentiary weight than a journalist’s inference from a press release, and collapsing the two produces a number no serious publication can cite.

Every entry below links to a primary source retrieved at entry time and is assigned an Attribution Tier. Entries where the company’s own documents do not mention AI are still listed — but they are marked Tier 3, and they are excluded from the documented-attribution total.

Headline figures (4 October 2026)

Measure	Value
Company entries logged	13
Benchmark entries logged	5
Employees in Tier 1 entries (regulatory filing)	21,000
Employees in Tier 2 entries (named executive, on record)	26,120
Documented-attribution subtotal (Tier 1 + Tier 2)	47,120
Employees in Tier 3 entries (press inference only)	28,459
Attribution Gap Ratio™ (AGR™)	0.604

Sixth reading. Twelve company entries, not a market-wide estimate. The ratio has risen for the first time. The AGR™ has moved 0.85 → 0.63 → 0.60 → 0.60 → 0.60 → 0.604 across six readings. Every previous move was downward and mechanical — Tier 2 entries entering the denominator. This one is the opposite: a Tier 3 entry entered the numerator while the denominator stood still.

The rise is small and rests on a single entry, so it is not yet a signal. But it is the direction the previous reading predicted would appear if AI-attributed cuts were being relabelled rather than actually declining, and it arrived alongside a second consecutive month of AI falling out of the top spot in Challenger’s data. Both are set out below. The prediction is now on the record and testable; it is not yet confirmed.

Two entries were added this cycle: Challenger’s September report, released 1 October, and the first WARN-sourced entry in the dataset — an Oracle filing in Washington state. The WARN addition closes a gap the maintenance protocol has flagged since the tracker opened.

The Attribution Tier framework

Every company entry is assigned exactly one tier, based on the strongest documentary evidence available at entry time.

Tier 1 — Documentary attribution. The company names AI as a cause of workforce reduction in a document filed with a securities regulator (10-K, 10-Q, 8-K, 20-F, or national equivalent). This is the highest standard: the language has passed corporate legal review and carries liability.

Tier 2 — Executive attribution on the record. A named executive attributes the reduction to AI in a shareholder letter, earnings call, published memo, or on-record interview. Attributable to a person, but not a regulated disclosure.

Tier 3 — Press attribution only. The reduction is real and sourced, but no company document or named executive connects it to AI. The AI framing originates with reporting, analysis, or a tracker. Logged for completeness; excluded from the documented-attribution total.

A tier can be upgraded later — several Tier 3 entries will become Tier 1 when the relevant annual filing is published. Tier changes are logged in the CSV changelog and never made silently.

Tracked entries — reverse chronological

1 October 2026 · Challenger: AI slips again, to fifth place — but still leads the year

Type: Benchmark dataset · Tier: 1 (originating organization’s own report)

AI was the fifth-most cited reason for September job cuts, with 3,961 — about 9% of the month’s total. That is a second consecutive month outside the top spot after five months leading, and it holds AI near the floor it hit in August rather than showing any rebound: 3,462 in August, 3,961 in September.

Market and Economic Conditions led September with 8,789 cuts (20%), reaching 114,124 year-to-date. Demand Downturn accounted for 6,515, its highest monthly total since February 2023. Restructuring took 6,243 for the month and 79,892 for the year. Total US announcements fell to 43,281, down 18% from August and the fewest for a September since 2022. Andy Challenger described employers as being in a wait-and-see period, citing high energy costs, an uncertain war in Iran, a possible rate hike and the likelihood of surging healthcare costs.

The year-to-date figure is the one that matters, and it cuts against the monthly headline. AI has been cited in 120,136 announcements in 2026, roughly 21% of all cuts, and remains the leading reason year-to-date despite two months off the monthly top spot. Any reading of these two months as “AI layoffs are over” is not supported by Challenger’s own annual column.

Revisiting the three readings posed on 26 September. Last cycle set out three explanations for August’s collapse and declined to choose. September narrows them, without settling them:

• Seasonal volatility is now the weakest. Two consecutive months at the floor is harder to attribute to a light announcement month than one.
• Genuine decline remains live, and the broader fall in total announcements supports it.
• Relabelling also remains live. The reasons that displaced AI — Market and Economic Conditions, Demand Downturn, Restructuring — are precisely the generic categories a company would use if it chose to stop naming AI.

Nothing here distinguishes the second from the third, and this tracker will not pretend otherwise. What would distinguish them is company-level evidence: if relabelling is happening, Tier 1 and Tier 2 additions should dry up while Tier 3 additions continue. That is exactly what happened this cycle — the only company entry added was Tier 3. One cycle is not a pattern, and this is recorded as an observation to test, not a finding.

Primary source: Challenger, Gray & Christmas, September 2026 Job Cut Report · Retrieved 4 October 2026

14 September 2026 · Oracle: 359 Washington jobs in a WARN filing that does not mention AI

Type: Company layoff · Tier: 3 · Employees: 359 · Flag: pending_primary

Oracle America filed a Worker Adjustment and Retraining Notification with the Washington State Employment Security Department on 14 September 2026, covering 359 workers at Oracle’s Seattle locations and remote staff across the state. The separation date is 13 November 2026. Oracle stated the layoffs are permanent and that the Seattle offices are not closing. Reporting puts Oracle’s cumulative Washington WARN layoffs at 1,112 since August 2025.

This is the first WARN-sourced entry in the dataset, closing a sourcing gap the maintenance protocol has carried since the tracker opened. State WARN notices establish the fact and the headcount with government-filing reliability — but they almost never establish attribution, and this one does not. Internal emails to affected employees reportedly cited a broader organizational change. The entry is therefore Tier 3.

Why this is Tier 3 when Oracle is also the tracker’s only Tier 1 entry. The 10-K language naming AI covers fiscal 2026, which ended 31 May 2026. This filing is from September. A company’s attribution in one period does not transfer to a reduction in another — if it did, every subsequent Oracle cut would inherit a Tier 1 label indefinitely. The tier travels with the document, not with the company.

No double counting, stated explicitly. This event falls outside fiscal 2026 and is therefore not inside the 21,000 net headcount change in the Oracle Tier 1 entry. The earlier 2026 WARN rounds — roughly 1,700 across California, Missouri and Washington, with separations between late May and early June — do fall inside that net change and are deliberately not logged as separate rows. Logging them would inflate the Oracle total by counting the same people twice.

Primary source: Washington State Employment Security Department, WARN notice filed 14 September 2026 · Flag: pending_primary Verification note: The figure is sourced to reporting that states it reviewed the notice directly. The ESD WARN record itself has not been retrieved and must be before this entry is treated as fully verified.

3 September 2026 · Challenger: the AI streak breaks — AI falls to the fourth-most cited reason, its lowest month since December 2025

Type: Benchmark dataset · Tier: 1 (originating organization’s own report)

This is the most significant reversal in the dataset since the tracker opened.

After leading all stated layoff reasons for five consecutive months, artificial intelligence fell to fourth place in August 2026, cited in just 3,462 cuts. That is the lowest monthly AI-attributed total since December 2025, when 142 cuts were attributed to the reason. Restructuring led the month instead, with 16,173 cuts, or 31% of the total.

For scale, the collapse is nearly elevenfold in three months: AI was cited in 38,579 cuts in May, 14,029 in June, 10,970 in July, and 3,462 in August.

US employers announced 52,881 job cuts in August, up 58% from July but the lowest August total since 2022; 529,914 year-to-date, down 41% from the same period in 2025. Technology announced 6,103 cuts in August, its lowest monthly total of the year, for a year-to-date total of 155,126 — still up 52% year on year and 29% of all 2026 cuts, more than any other sector. Andy Challenger noted that while companies are planning to hire more than last year, those positions do not appear to be filling quickly.

What this does and does not mean, stated carefully. A single month is not a trend, and there are at least three readings of the drop that the data cannot distinguish between:

1. Companies genuinely attributed fewer cuts to AI in August.
2. Companies made the same kinds of cuts but described them differently — note that Restructuring rose to first place in the same month.
3. August is seasonally light for announcements, and the AI-attributed subset is volatile at low volumes.

Reading 2 is the one this tracker was built to detect, and it is consistent with Challenger’s own July warning that companies would become more careful in their announcements as regulation takes shape. If AI cuts are being relabelled as restructuring rather than actually declining, the Tier 1 and Tier 2 pools shrink while the underlying reality does not — and the AGR™ would be expected to rise in subsequent readings as press attribution continues where company attribution stops.

No conclusion is drawn here. The September and October reports are what will separate the three readings, and this entry will be revisited when they publish.

Primary source: Challenger, Gray & Christmas, August 2026 Job Cut Report · Retrieved 26 September 2026

6 August 2026 · Challenger: AI leads stated layoff reasons for a fifth consecutive month, as total cuts fall to a two-year low

Type: Benchmark dataset · Tier: 1 (originating organization’s own report)

US employers announced 33,429 job cuts in July 2026 — the lowest monthly total in two years, down 27% from June and 46% below July 2025. Artificial intelligence nevertheless led all stated reasons for the fifth consecutive month, cited in 10,970 cuts, or 33% of the month’s total. Year-to-date, AI has been cited in 112,713 announcements, roughly 24% of all cuts; cumulatively since 2023 the figure stands at 184,538. Technology announced 9,867 July cuts for a 2026 total of 149,023, up 67% year on year and 31% of all cuts.

Andy Challenger’s framing is worth recording precisely, because it cuts against the alarmist reading of his own data: hiring is up 25% on last year, so while AI is shifting the labour market, it is not dismantling it.

The report also concedes the problem this tracker exists to address. Challenger notes there remains genuine ambiguity about what counts as an AI-related cut — some employers cite AI explicitly, while others reference new technology deployments and allude to AI without linking it to the reductions — which is why the firm tracks those announcements in a separate category. Challenger further warned that as regulation takes shape, companies will become more careful in their announcements, which would make the impact of AI on jobs harder to track, not easier.

That last point has a direct consequence for this tracker: the Tier 1 pool may shrink over time not because AI is causing fewer job losses, but because fewer companies will be willing to write it down.

Primary source: Challenger, Gray & Christmas, July 2026 Job Cut Report · Retrieved 13 August 2026

22 July 2026 · Monday.com: 20% cut in a restructuring branded as a pivot to an AI Work Platform

Type: Company layoff · Tier: 2 · Employees: 620 (≈20%) · Flag: source_inconsistency

Monday.com cut roughly 20% of its global workforce, restructuring around what it described as its AI Work Platform. The language was deliberately framed as a pivot rather than a retreat. Shares rose 2.3% after the announcement.

Source inconsistency, published not resolved: TechChannel reports approximately 620 employees; layoffs.fyi lists 630 for the same event. The lower figure is used per counting conventions pending the company’s own disclosure.

Primary source: TechChannel — enterprise software layoffs analysis · Retrieved 24 August 2026

6 July 2026 · Microsoft: 4,800 roles cut in a memo that names AI and denies AI replacement in the same breath

Type: Company layoff · Tier: 2 (indirect attribution) · Employees: 4,800 (2.1%) · Flag: methodology_review

Microsoft eliminated approximately 4,800 roles, about 2.1% of its global workforce. EVP and Chief People Officer Amy Coleman published the memo on the Official Microsoft Blog, framing the reduction as a refocusing of people and investments for a fast-changing industry. She then wrote, directly and unusually: the roles eliminated are not being replaced by AI — while adding that AI is changing how work gets done.

Roughly 1,600 of the cuts fall in Xbox, with a further 1,600 planned through fiscal 2027; Xbox’s difficulties were attributed in part to Game Pass growth falling short of expectations, which is not an AI story. Microsoft also redeployed more than 4,000 employees over the preceding year, reassigned around 500 more, and reported that over 30% of eligible employees took voluntary retirement.

This is the hardest tier call in the dataset, and it is worth being explicit about the reasoning. A company’s own published document links the restructuring to AI-driven industry change — that is a genuine company attribution, and it satisfies Tier 2. But the same executive explicitly denies that AI is replacing the eliminated roles. Both things are true at once.

The entry is therefore classified Tier 2 with indirect attribution, and carries a standing caveat: it must never be cited as an instance of AI replacing workers. Microsoft said the opposite, in writing, on its own blog.

Every competing tracker lists this event as an AI layoff without the caveat. That is the AI-washing problem operating in the other direction — not a company overstating AI, but trackers overstating a company that went out of its way to be precise.

Methodology note: this entry establishes a precedent for indirect attribution under Tier 2 and requires Editorial Director sign-off before publication. If the precedent is rejected, the entry moves to Tier 3 and the AGR™ recomputes to 0.70.

Primary source: Amy Coleman, “The latest in our company transformation,” Official Microsoft Blog, 6 July 2026 · Retrieved 13 August 2026 Verification note: The blog URL returned a redirect loop to our fetcher on 13 August. Memo language is confirmed identically across ABC News, Fox Business, Windows Central and TechPowerUp, all quoting the published post directly. Direct retrieval should be reattempted at the next sweep.

2 July 2026 · Challenger, Gray & Christmas: AI leads stated layoff reasons for a fourth consecutive month

Type: Benchmark dataset · Tier: 1 (originating organization’s own report)

Challenger’s June report recorded 45,849 announced US job cuts, down 53% from May. Artificial intelligence accounted for 14,029 of them, or 31% of the month’s total — the fourth consecutive month AI ranked as the single most-cited reason, a streak with no precedent in the firm’s data. Year-to-date through June, AI had been cited in 101,743 announced cuts, approximately 23% of all US layoff announcements. Cumulatively since 2023, when Challenger first tracked AI as a distinct reason, the figure stands at 173,568.

Challenger’s methodology counts stated reasons in announcements, which is closer to Axis Tier 2 than Tier 1 and is not restricted to the technology sector. It is the most rigorous independent benchmark available and is used here for context, not as a component of the Axis totals.

Primary source: Challenger, Gray & Christmas, June 2026 Job Cut Report · Retrieved 1 August 2026

22 June 2026 · Oracle: first major technology company to name AI as a workforce-reduction cause in a 10-K

Type: Company layoff · Tier: 1 · Employees: 21,000 (net headcount change, Axis-calculated)

Oracle’s Form 10-K for the fiscal year ended 31 May 2026 discloses approximately 141,000 full-time employees, of whom roughly 49,000 are in the United States and 92,000 international. Under the risk factor headed “Our periodic workforce restructurings and reorganizations can be disruptive,” the filing states that the adoption and deployment of AI technologies across Oracle’s operations “have resulted, and may continue to result, in reductions to our workforce.”

This is the strongest attribution standard currently available anywhere in the dataset: language in a securities disclosure, reviewed by counsel, describing a material risk.

Additional figures from the same filing: restructuring costs of approximately $1.84bn for the year against revenue of $67.4bn, with free cash flow at negative $24bn.

Two caveats that most coverage omits, and that matter for anyone citing this figure:

1. The 21,000 is a calculated net change, not a disclosed layoff number. Oracle’s FY2026 filing reports 141,000 employees; the FY2025 filing reported approximately 162,000. The difference is Axis-calculated from two filings and includes attrition, divestitures and non-replacement, not only involuntary separations. Oracle did not publish a layoff total.
2. The filing does not quantify the AI-attributable portion. The same risk factor attributes restructuring to management changes, product changes, performance issues, strategic changes and acquisitions, and then adds AI as a further factor. Treating all 21,000 as AI-caused is not supported by the document.

Widely circulated figures of 30,000 cuts or an 18% reduction do not reconcile with the filed headcount and are not used here.

Primary source: Oracle Corporation, Form 10-K, FY ended 31 May 2026 — SEC EDGAR · Retrieved 1 August 2026

4 June 2026 · Challenger: May sets the highest monthly AI-attributed total on record

Type: Benchmark dataset · Tier: 1 · Flag: pending_primary (release date)

May 2026 produced 38,579 announced cuts citing AI — the highest monthly total ever recorded for that reason since Challenger began tracking it in 2023, and the third straight month AI led all stated reasons. Andy Challenger, the firm’s chief revenue officer, tied the surge to companies restructuring aggressively as they reposition for an AI-driven economy, alongside a sharp rise in cuts linked to mergers and bankruptcies.

For scale: this single month exceeds the firm’s entire 2024 AI-attributed count.

Primary source: Challenger, Gray & Christmas, May 2026 Job Cut Report (PDF) · Retrieved 1 August 2026 Verification note: The report is confirmed; the exact press-release date (inferred as the first Thursday of June, per Challenger’s standard schedule) requires confirmation against the firm’s press page.

20 May 2026 · Intuit: 3,000 cut in a restructuring built around infusing AI across its services

Type: Company layoff · Tier: 2 · Employees: 3,000 (17%) · Flag: pending_primary

Intuit eliminated roughly 3,000 roles, about 17% of its workforce, in a restructuring reported as centred on reducing complexity and reallocating resources toward AI. CEO Sasan Goodarzi told staff the company was reducing complexity and simplifying its structure in order to deliver better products.

Tier strengthened, 24 August 2026. This entry previously carried a caveat that the CEO quote did not itself name AI. Reuters reports that Goodarzi sent a memo to employees stating the layoffs would support the company’s effort to infuse AI across its services — a direct AI attribution by a named executive in a company communication. The caveat is withdrawn; the entry is now a standard Tier 2. Direct retrieval of the memo remains outstanding.

Source withdrawn, 26 September 2026. This entry was previously cited to a competing layoff compilation. Axis does not cite or link competing compilations — the tracker computes its own figures from primary documents. The citation is withdrawn and the entry flagged compilation_sourced until a primary or directly-quoting source is retrieved. The underlying facts are unchanged; only the citation is. Resolution path: Intuit investor relations and any Form 8-K accompanying the restructuring announcement.

20 May 2026 · Meta: 8,000 cut as Zuckerberg frames the reduction around AI competition

Type: Company layoff · Tier: 2 · Employees: 8,000 (≈10% of workforce)

Meta began notifying approximately 8,000 employees on 20 May, roughly 10% of its workforce. In an internal memo, CEO Mark Zuckerberg told staff that “AI is the most consequential technology of our lifetimes” and that the companies leading it will define the next generation, framing the reduction as competitive necessity rather than cost control. Separately, he warned that “success isn’t a given.” The restructuring also left roughly 6,000 open positions unfilled and moved about 7,000 employees into AI-focused roles.

The reorganization has since been publicly qualified by its own author: at a July town hall Zuckerberg reportedly told employees that AI agent development had not accelerated as expected and that the restructuring had not been as clean as planned.

Tier 2, not Tier 1: the memo was internal and has not been published by Meta. It is quoted by multiple outlets that state they viewed it directly, and its authenticity was confirmed to NBC News by a source familiar with the matter.

Primary source: CNBC — Zuckerberg’s memo on the Meta layoffs · Retrieved 1 August 2026 Resolution path to Tier 1: Meta Q2 2026 Form 10-Q headcount disclosure and any restructuring-related 8-K (CIK 0001326801).

13 May 2026 · Cisco: 4,000 cut, with the CFO stating the restructure was not about savings

Type: Company layoff · Tier: 2 · Employees: 4,000 (5%) · Flag: source_inconsistency

Cisco announced nearly 4,000 job cuts, about 5% of its workforce, despite reporting better-than-expected profit and revenue. CFO Mark Patterson said the move was not a savings-driven restructure, describing it instead as realigning resources around silicon, optics, security and AI. Shares reportedly rose 17% in after-hours trading.

This is a clean Tier 2: a named executive, on the record, naming AI as part of the rationale — and notably ruling out the cost-cutting explanation that would otherwise be the default reading of a profitable company cutting staff.

Source inconsistency, published rather than resolved silently: TechCrunch dates the announcement 14 May 2026; layoffs.fyi lists 13 May 2026. Cisco’s own release is required to settle it. The entry uses 14 May pending that retrieval.

Source withdrawn, 26 September 2026. This entry was previously cited to a competing layoff compilation. Axis does not cite or link competing compilations — the tracker computes its own figures from primary documents. The citation is withdrawn and the entry flagged compilation_sourced until a primary or directly-quoting source is retrieved. The underlying facts are unchanged; only the citation is.

7 May 2026 · Cloudflare: 20% of staff cut, with no AI attribution from the company

Type: Company layoff · Tier: 3 · Employees: ~1,100 (≈20%) · Flag: pending_primary (date, headcount)

Cloudflare reduced headcount by approximately 20% despite growth exceeding 30%. In a note posted to X, CEO Matthew Prince said the cuts were made “because business is changing, and to win the future, Cloudflare needs to change with it.”

That statement is on the record and attributable to a named executive — but it does not mention AI, automation, or machine learning. Competing trackers list Cloudflare in their AI layoff tables. On this tracker’s criteria, an executive declining to name AI is not an AI attribution, so the entry is Tier 3.

Correction, 13 August 2026: this entry was previously dated 21 May 2026. That was an Axis error. TechCrunch places the announcement on 7–8 May, corroborated by the layoffs.fyi row dated 7 May. The date has been corrected and the change logged in the CSV.

Source withdrawn, 26 September 2026. This entry was previously cited to a competing layoff compilation. Axis does not cite or link competing compilations — the tracker computes its own figures from primary documents. The citation is withdrawn and the entry flagged compilation_sourced until a primary or directly-quoting source is retrieved. The underlying facts are unchanged; only the citation is. Verification note: The headcount figure (~1,100) still requires confirmation against Cloudflare’s investor relations disclosures and Prince’s account directly.

5 May 2026 · Coinbase: 700 cut, with AI named alongside market conditions

Type: Company layoff · Tier: 2 · Employees: 700 (14%) · Flag: pending_primary

Coinbase cut approximately 700 employees, around 14% of staff. CEO Brian Armstrong cited both current market conditions and AI changing how the company works; a second account describes the cuts as driven partly by Coinbase needing to become AI-native.

Mixed attribution. AI is named by the CEO, which satisfies Tier 2 — but it is named alongside market conditions, so the AI-attributable portion is not isolable. This is the common shape of real-world attribution and the reason ai_attributed_portion is recorded as not_quantified rather than assumed.

Source withdrawn, 26 September 2026. This entry was previously cited to a competing layoff compilation. Axis does not cite or link competing compilations — the tracker computes its own figures from primary documents. The citation is withdrawn and the entry flagged compilation_sourced until a primary or directly-quoting source is retrieved. The underlying facts are unchanged; only the citation is.

15 April 2026 · Snap: 1,000 cut, with the CEO citing AI-enabled small teams

Type: Company layoff · Tier: 2 · Employees: 1,000 (16%) · Flag: pending_primary

Snap cut approximately 1,000 roles, about 16% of its workforce. CEO Evan Spiegel said the company had already seen small squads using AI tools drive meaningful progress across several important initiatives, with AI used to streamline operations and support a smaller workforce.

A clean named-executive attribution: AI is explicitly connected to the company operating with fewer people.

Source withdrawn, 26 September 2026. This entry was previously cited to a competing layoff compilation. Axis does not cite or link competing compilations — the tracker computes its own figures from primary documents. The citation is withdrawn and the entry flagged compilation_sourced until a primary or directly-quoting source is retrieved. The underlying facts are unchanged; only the citation is.

16 March 2026 · Dell: 11,000 fewer employees in the FY2026 10-K — and no mention of AI

Type: Company layoff · Tier: 3 · Employees: 11,000 (≈10.2%)

Dell Technologies’ annual report for the fiscal year ended 30 January 2026 discloses approximately 97,000 employees, down from about 108,000 a year earlier. The filing attributes the reduction to “disciplined cost management in coordination with our ongoing business modernization initiatives” alongside employee reorganizations, restricted external hiring and facility consolidation. Severance charges totalled $569 million, down from $693 million in FY2025 and $648 million in FY2024.

This entry is Tier 3 despite being sourced from a 10-K, and the reason is the whole point of the framework. A regulatory filing is only Tier 1 evidence if it actually attributes the reduction to AI. Dell’s does not. Moreover, this is the third consecutive fiscal year of roughly 10% reduction — headcount has fallen from about 133,000 in FY2023 — a pattern established before the current AI capital expenditure cycle. Dell is simultaneously expanding its AI server business, which is why the cuts are widely reported as AI-driven; that is an inference, not a disclosure.

Competing trackers list Dell’s 11,000 as an AI layoff. On the evidence in Dell’s own filing, that attribution is unsupported.

Primary source: Dell Technologies Inc., Form 10-K, FY ended 30 January 2026 — SEC EDGAR, CIK 0001571996 · Flag: pending_primary Verification note: Filing language confirmed via multiple independent outlets quoting the document identically. The direct EDGAR accession URL has not yet been retrieved and must be fetched before this entry is treated as fully verified.

26 February 2026 · Block: 4,000+ cut, with “intelligence tools” named as the stated cause

Type: Company layoff · Tier: 2 · Employees: 4,000+ (≈40%)

Block announced alongside Q4 2025 results that it would reduce headcount from more than 10,000 to just under 6,000. In a letter to shareholders, CEO Jack Dorsey wrote that “Intelligence tools have changed what it means to build and run a company” and that a significantly smaller team using those tools can do more and do it better. He further predicted that most companies would reach the same conclusion and make similar structural changes within a year, adding that he would rather arrive there on Block’s own terms than reactively.

This is the most explicit CEO-level AI attribution in the dataset. It is also the entry where the AI-washing question is sharpest: Dorsey acknowledged the cuts were partly a response to pandemic-era overhiring, and the announcement accompanied record gross profit growth. Block’s shares rose more than 20%.

Tier 2, not Tier 1: the shareholder letter is an investor communication. It is customarily furnished as an exhibit to a Form 8-K, which would move this entry to Tier 1 on verification.

Primary source: Block Q4 2025 shareholder letter, as reported by CNN Business · Retrieved 1 August 2026 Resolution path to Tier 1: Block Inc. Form 8-K filed on or about 26 February 2026, EDGAR CIK 0001512673, Item 2.02 exhibit.

28 January 2026 · Amazon: 16,000 roles cut — and the memo does not mention AI

Type: Company layoff · Tier: 3 · Employees: 16,000

Amazon announced approximately 16,000 role reductions, its second large corporate round in three months. In her memo to employees, SVP of People Experience and Technology Beth Galetti said the reductions would strengthen the organization by “reducing layers, increasing ownership, and removing bureaucracy,” and that most US-based employees would have 90 days to find an internal role.

AI is not cited in the January memo. This is verifiable and was noted at the time by NBC News. The AI framing in nearly every tracker listing this event comes from context — Amazon’s $125bn capital expenditure guidance, and CEO Andy Jassy’s June 2025 statement that AI efficiency gains would likely shrink corporate headcount — not from the announcement itself.

The contrast within Amazon is instructive. The October 2025 round of 14,000 corporate cuts was accompanied by a Galetti memo that explicitly invoked AI as the most transformative technology since the internet. Same company, same executive, three months apart: one round qualifies as Tier 2, the other does not. A tracker that cannot represent that distinction is not measuring attribution at all.

Primary source: NBC News — Amazon cuts 16,000 jobs (quoting the Galetti memo) · Retrieved 1 August 2026 Resolution path: Amazon publishes these memos on its corporate news site; the original post should be linked directly. October 2025 round to be logged as a separate Tier 2 entry once its exact date is confirmed.

Why the attribution gap exists: markets reward the AI framing

The AGR™ measures the gap. This is the mechanism behind it, and it is the strongest argument for why a tracker like this one is necessary.

Companies that frame workforce reductions as AI-driven restructuring have been rewarded with higher valuations. Cisco’s shares jumped 17% in after-hours trading on its restructuring announcement. Monday.com’s stock rose 2.3% after cutting 20% of its workforce. ServiceNow, whose cuts coincided with its AI portfolio crossing $1bn in annual contract value, saw shares climb roughly 9% over the following week. Block rose more than 20%.

There is a straightforward incentive here: describing a layoff as an AI pivot converts a cost signal into a growth signal. That does not mean any given company is being dishonest — Cisco’s CFO went out of his way to say the restructure was not savings-driven, and Microsoft went out of its way to deny AI replacement entirely. But it does mean the AI framing carries a financial reward, and any measurement built on self-reported attribution has to account for that.

It cuts both ways, which is the part most commentary misses. Tier 1 and Tier 2 counts are inflated where AI is invoked because investors respond to it. They are also suppressed where companies avoid the word — Challenger has warned that as regulation takes shape, companies will become more careful in their announcements, making AI’s impact on jobs harder to track, not easier.

The tier framework does not resolve this. It makes it visible, which is the most an honest tracker can do.

A named executive has conceded the problem

The AI-washing argument no longer rests on inference from share-price reactions. In February 2026, OpenAI CEO Sam Altman stated publicly that there is some AI washing where people blame AI for layoffs they would otherwise have carried out.

That is a principal in the AI industry — with every commercial incentive to maximise perceptions of AI’s labour impact — conceding that the attribution is sometimes false. It does not tell us which attributions are false, which is precisely what the tier framework is for. But it removes the question of whether the phenomenon exists.

Regulation may force the issue. California’s proposed SB 951, the Worker Technological Displacement Act, would require 90 days’ advance notice before AI-driven layoffs and disclosure of the specific AI systems involved. The bill remains pending and this tracker does not log pending legislation as fact. If enacted, it would create a disclosure obligation that moves AI attribution from voluntary framing into regulated filing — which would substantially expand the Tier 1 pool and make the AGR™ measurable against a legal standard rather than a voluntary one.

The near-term effect could run the other way. A company facing a 90-day notice requirement triggered by naming AI has an obvious reason not to name it.

How we curate this list

The inclusion criteria are the methodology. If you cite a number from this tracker, this section is what you are citing.

What counts as an entry

An event is logged when all of the following are true:

1. A workforce reduction is announced, disclosed, or documented by an identifiable organization.
2. The event has a specific date.
3. A primary or directly-quoting source is retrieved and linked at entry time.
4. Either the company connects the reduction to AI (Tier 1 or 2), or the event is widely attributed to AI by other trackers and press while the company’s own documents do not make that connection (Tier 3).

Criterion 4 is deliberate. Tier 3 entries exist so that the gap between claimed and documented attribution is measurable rather than invisible.

What does not count

• Hiring freezes and unfilled roles. Meta’s 6,000 unfilled positions are noted in context but not counted as employees cut.
• Redeployments. Meta’s ~7,000 employees moved into AI roles are noted, not counted.
• Attrition-only headcount declines with no announced reduction and no filing language.
• Forecasts and multi-year targets until the reduction is actually executed and disclosed. A stated intent to cut 20,000 “by 2027” is not a dateable event.
• Rumours, unnamed-source reports, and analyst estimates presented without a company document or named executive.
• Aggregate projections. This tracker logs discrete dated events. Statistical modelling of AI’s labour-market effect belongs elsewhere.

Counting conventions

• Employees is the figure in the primary source. Where the source gives a range, the lower bound is used.
• Net headcount changes calculated from two filings are marked axis_calculated=yes and are never presented as layoff counts.
• Percentages are reproduced only when stated in the source. They are never derived by Axis from an assumed denominator.
• Benchmark entries (Challenger and equivalent) are flagged event_type=benchmark and are excluded from all Axis totals to prevent double counting.
• Tier assignment reflects the strongest evidence at entry time and is re-examined whenever a new filing is published.

The Attribution Gap Ratio™ (AGR™)

Formula:

AGR™ = Σ(employees in Tier 3 company entries) ÷ Σ(employees in Tier 1 + Tier 2 company entries)

Benchmark entries are excluded. Entries with no numeric employee figure are excluded from both sides and noted in the CSV.

Computation, 4 October 2026:

Tier 1:  Oracle (FY2026 10-K) 21,000          = 21,000

Tier 2:  Meta 8,000 + Microsoft 4,800
         + Cisco 4,000 + Block 4,000
         + Intuit 3,000 + Snap 1,000
         + Coinbase 700 + Monday.com 620      = 26,120

         Documented-attribution subtotal      = 47,120

Tier 3:  Amazon 16,000 + Dell 11,000
         + Cloudflare 1,100
         + Oracle (Sept WARN) 359             = 28,459

AGR™ = 28,459 ÷ 47,120 = 0.604

Reading it: for every 100 jobs cut where the company itself documented AI as a cause, a further 60 are counted as “AI layoffs” elsewhere without any company attribution behind them.

This reading is published to three decimal places, deliberately. At two decimals the ratio still rounds to 0.60 and the first upward movement in the series would be invisible. Rounding that hides a direction change is a presentation choice, not a neutral one. Subsequent readings will carry three decimals until the series moves enough that two suffice.

Reading the movement correctly. The ratio has moved 0.85 → 0.63 → 0.60 → 0.60 → 0.60 → 0.604 since 1 August. The five earlier readings fell or held flat for mechanical reasons: Tier 2 entries entering the denominator, then no entries at all. This reading rose, and for the opposite reason — a Tier 3 entry entered the numerator while the denominator did not move.

One entry is not a trend, and the movement is three thousandths. What makes it worth flagging is that it is the first directional change in the series and it matches a prediction made in the previous cycle. It is recorded as such and nothing more. This is not evidence that attribution practice improved in three weeks. Tier 2 entries were added at each reading and the Tier 3 total was unchanged, so the denominator grew and the ratio fell mechanically. Until the entry set stabilises, AGR™ movement reflects coverage growth more than it reflects the market.

What it is not: the AGR™ is not a measure of dishonesty, and a Tier 3 classification is not an accusation. Companies have legitimate reasons not to attribute cuts to AI, and the press has legitimate reasons to infer a connection. The ratio measures the distance between the two — nothing more. Every input is in the CSV and every figure is reproducible.

Dataset

The full dataset is published as CSV under CC BY 4.0. Attribution: Axis Intelligence Research, AI Layoff Tracker.

Columns: entry_id, event_date, organization, event_type, attribution_tier, employees, pct_workforce, ai_attributed_portion, source_org, source_document, primary_doc_type, source_url, retrieved_date, is_primary, axis_calculated, method_note, verification_flag.

Download the CSV →

Open verification flags

We publish these rather than resolving them silently.

Entry	Flag	Resolution path
Microsoft	methodology_review	Editorial Director sign-off required on the indirect-attribution Tier 2 precedent. If rejected, entry moves to Tier 3 and AGR™ recomputes to 0.79
Oracle (Sept WARN)	pending_primary	Retrieve the Washington ESD WARN record itself; figure currently rests on reporting that states it reviewed the notice
Monday.com	compilation_sourced	Citation withdrawn 26 Sep. Retrieve Monday.com IR communication or the restructuring announcement
Intuit	compilation_sourced	Citation withdrawn 26 Sep. Retrieve the Goodarzi memo directly via Intuit IR
Coinbase	compilation_sourced	Citation withdrawn 26 Sep. Retrieve Armstrong’s communication via Coinbase IR
Snap	compilation_sourced	Citation withdrawn 26 Sep. Retrieve Spiegel’s statement via Snap IR
Cloudflare	pending_primary	Headcount (~1,100) via Cloudflare IR; original Prince post on X
Dell FY2026	pending_primary	Retrieve EDGAR accession URL, CIK 0001571996, 10-K for FY ended 30 Jan 2026
Block	pending_primary	Locate Form 8-K exhibit containing the shareholder letter, CIK 0001512673
Challenger May report	pending_primary	Confirm press-release date on challengergray.com
Amazon Oct 2025	pending_entry	Confirm exact date; log as separate Tier 2 entry
Meta	tier_review	Upgrade to Tier 1 if Q2 2026 10-Q quantifies the reduction

Closed since 1 August: Cloudflare event date (corrected to 7 May 2026); Challenger June report date (confirmed as 2 July 2026); Intuit tier caveat (withdrawn 24 August after Reuters confirmed the Goodarzi memo names AI); Cisco date and CFO name (resolved 26 September via the Form 8-K — date corrected from 14 May to 13 May, CFO confirmed as Mark Patterson, entry re-sourced to SEC EDGAR).

WARN integration opened, 4 October 2026. The Oracle Washington filing is the first WARN-sourced entry. State WARN notices establish fact and headcount reliably but rarely establish attribution, so they will typically produce Tier 3 entries. They are nonetheless the strongest available check on employee counts and will be swept systematically from this cycle forward.

Sourcing correction, 26 September 2026. Four entries — Monday.com, Intuit, Coinbase and Snap — were previously cited to competing layoff compilations. Axis does not cite or link competing compilations; the tracker exists to compute its own figures from primary documents, and sourcing a row to another tracker undermines the entire premise. All four citations have been withdrawn and the entries flagged compilation_sourced.

The facts in those four entries are unchanged and the entries remain in the dataset at their existing tiers. What has been withdrawn is the citation, not the finding. This is deliberately more visible than quietly swapping in a different link: the rows are marked as inadequately sourced until a primary or directly-quoting document is retrieved, and anyone citing them should know that.

Entry removed, 24 August 2026. A row for Wix.com (1,000 employees, 25 May 2026) was present in an earlier build of the dataset. Its only source was a content aggregator, which does not meet any tier of this tracker’s source hierarchy. The row has been removed rather than downgraded, because an entry that was never sourced to a primary or directly-quoting document should not have been logged at all. Entry IDs were renumbered in the same pass; anyone who downloaded the CSV before this date should re-download.

Frequently asked questions

Why is this AI layoff total lower than other trackers?

Because the totals are measuring different things. The documented-attribution subtotal counts only cuts where the company itself named AI in a filing or through a named executive. Trackers reporting figures several times higher are including press-inferred attributions, which appear here as Tier 3 and are reported separately rather than merged in.

What are the AI layoff attribution tiers?

Tier 1 is documentary attribution, where a company names AI as a cause of workforce reduction in a securities filing. Tier 2 is executive attribution, where a named executive makes the connection in a shareholder letter, earnings call, published memo or on-record interview. Tier 3 covers reductions that are real and sourced but where no company document or named executive connects them to AI; the attribution originates with reporting. Tier 3 entries are excluded from the documented-attribution total.

A Tier 3 classification is not second-guessing the reporting. Those entries are logged, sourced and counted — simply in their own column. The reporting may well be correct; the point is that a reader can see which figures rest on a company document and which rest on an inference.

Does a Tier 1 classification mean AI caused all of those job losses?

No, and Oracle is the clearest example. Its filing names AI as a factor in workforce reductions without quantifying the portion. Tier 1 describes the quality of the attribution, not the completeness of the causation.

What is the Attribution Gap Ratio™?

It is the total employees in Tier 3 entries divided by the total employees in Tier 1 and Tier 2 entries. It measures the distance between AI job cuts that companies have documented and AI job cuts that are attributed by other parties. As of 4 October 2026 the ratio is 0.604, meaning that for every 100 jobs cut with documented company attribution, a further 60 are counted as AI layoffs elsewhere without company attribution behind them. It has moved 0.85 → 0.63 → 0.60 → 0.60 → 0.60 → 0.604 across six readings; each fall reflects Tier 2 entries being added to the denominator, not any reclassification of existing entries.

How often is the AI Layoff Tracker updated?

Weekly, on Fridays, and within 48 hours of any qualifying regulatory filing. The dateModified field is bumped on every update and the CSV is regenerated in the same pass.

Can the AI Layoff Tracker dataset be reused?

Yes. The full dataset is published as CSV under CC BY 4.0. Cite as: Axis Intelligence Research, AI Layoff Tracker, accessed [date]. If you are citing the documented-attribution subtotal, please state the tier basis — it is what makes the figure defensible.

Something is wrong or missing — how do I report it?

Send the primary document. Corrections backed by a source are applied within one business day and logged in the CSV changelog.

---

Others Sources

• Tech Layoffs Statistics
• AI Job Loss Statistics
• AI Job Displacement Statistics
• AI Replacing Jobs Statistics
• Digital Nomad Statistics
• Gig Economy Statistics