AI Layoffs Show In Job Listings First

• Job postings drop before layoffs do. Across four companies in early 2026, active listings fell off weeks to months before the cuts were announced. This job posting data is an early warning, not a record.

• Block: Listings fell ~80% in a late-February cliff. On Feb 26, 2026, Block cut ~40% of staff (10,000 → under 6,000), with Dorsey naming AI directly; the stock rose 20%+ that day.

• Oracle: Engineering listings collapsed from ~4,500 to under 800 in late February — about a month before the March 31 termination emails (~20,000–30,000 roles). Headcount fell ~13% year over year. The cuts fund a ~$50B AI buildout, making Oracle both the wave's biggest winner and a cutter of its own engineers.

• Coinbase: A slow slide, not a cliff — listings fell from ~580 to ~110 across late 2025 before the May 5, 2026 cut (~700 people, ~14%). AI plus a crypto downturn both played a role.

• Sprout Social: Listings wound down from ~60 to ~5 over a year before a 20% cut in July 2026 — and notably, it cut while beating guidance, restructuring toward AI from a position of strength.

• Which jobs go first: the labor market data shows a consistent order. Creative and sales roles get trimmed early because they're easy to cut; engineering gets cut deepest because AI coding tools mean fewer engineers do the same work; middle management thins in between.

• Next: where AI is actually adding jobs, and what the job market trends show about where headcount is heading instead.

AI-driven labor cuts, seen first in the job data

The clearest evidence that AI is reshaping hiring is visible in job postings first, which start moving before layoff notices go out. Across four example companies in the first half of 2026, the sequence repeats: active listings fall off, then weeks or months later the company announces cuts and names AI as the reason. Read in order, the charts are less a record of what happened than an early warning of it — one that shows up well before the job market news does.

Block: active listings fell roughly 80% in a single February cliff

The Block chart shows two purge-and-rebuild job posting cycles. Active listings dropped from about 330 to about 125 in a first cliff around March 2025, rebuilt across the year to roughly 330 by January 2026, then fell off a second, deeper cliff in late February—from about 330 to about 65, close to an 80% cut—before recovering toward 230. 

On February 26, 2026, Block cut about 40% of its workforce, more than 4,000 people, taking headcount from roughly 10,000 to under 6,000. Chief Jack Dorsey tied the decision directly to AI in a letter to shareholders, and the company credited its internal AI platform with a sharp rise in code shipped per engineer; the stock rose more than 20% the same day. 

Oracle: engineering-heavy listings collapsed from about 4,500 to under 800 in late February — roughly a month before the termination emails.

Total openings at Oracle ran above 6,000 in early 2025, rebuilt to a peak near 5,300 around the turn of 2026, then fell off a cliff in late February, cratering below 800 and bottoming in the 500–700 range by April before a partial rebuild toward 2,100. Now line that up against their recent layoff announcement. 

Oracle sent termination emails the morning of March 31, 2026, with no prior warning, in what reporting put at roughly 20,000 to 30,000 roles. Reuters, citing Bloomberg, had flagged the planned cuts on March 5.. The listings cliff came in late February, ahead of both. Oracle's annual filing later confirmed headcount had fallen to 141,000 from 162,000, an almost 13% reduction, and stated that deploying AI across its operations has resulted, and may continue to result, in workforce reductions. 

The cuts were explicitly tied to funding an aggressive AI-infrastructure buildout: roughly $50 billion in capital spending for the fiscal year. Oracle is simultaneously the buildout's biggest winner, so the accurate headline isn't "AI destroyed Oracle jobs" but that even AI's largest infrastructure beneficiary is cutting its software-engineering hiring to pay for the bet.

Coinbase: engineering listings slid from about 580 to roughly 110 across late 2025 and into 2026 — months before the cut was announced.

Coinbase registers more as a slope of reduced hiring, not a cliff. Computer & Mathematical carries the chart, peaking near 580 in late June 2025, then stepping down through the second half of the year to a trough around 110 before a modest rebuild toward 150. 

A layoff announcement came on May 5, 2026: about 700 people, roughly 14% of staff, completed in the second quarter. Brian Armstrong cited AI alongside a crypto downturn, describing a rebuild into a lean, AI-native org with no pure managers and, in some cases, one-person "pods" running fleets of agents. A Mizuho analyst argued the crypto winter is the real driver and AI the convenient explanation, a fair read given Coinbase's fourth-quarter loss, and both can be true at once. A down market forces the cut; AI pressurizes which roles don't come back. Either way, the pullback was visible in the listings months before the press release.

Sprout Social: listings wound down from about 60 to near 5 over a year.

Sprout Social: New Job Listings from January, 2025 to present

Sprout Social: Occupation shift away from Sales Reps and Financial Roles

Sprout's occupation chart peaks near 60 in mid-July 2025 and declines steadily to about 5 by mid-2026, an effective freeze, with Sales & Related thinning out earliest and Computer & Mathematical and Management following. The board approved the reduction on July 8, 2026, and notifications began the week of July 14: about 260 roles, roughly 20% of staff, the company's third layoffs round since 2024, tied explicitly to reallocating toward AI-powered social intelligence. What makes Sprout the cleanest "position of strength" example is that it cut 20% while beating guidance, restructuring toward AI, not retrenching from weakness. The openings, again, were gone long before the announcement made it official.

Which jobs go first

Creative and sales roles go first because they're easy to cut; at Coinbase, Arts & Design starts shrinking before the bigger reductions; at Sprout, Sales & Related drops ahead of the deeper cuts, a useful warning sign at companies that haven't announced AI layoffs yet. Engineering gets cut deepest: Computer & Mathematical is the largest category and the largest cut at both Coinbase and Oracle, because AI coding tools let each engineer do more, so companies need fewer of them. Block's own platform raised code output per engineer, Coinbase required AI assistants and saw non-technical staff ship code, and Oracle is automating engineering work directly. Middle management thins too: Coinbase cut "pure managers," Oracle cut senior technical staff. The pattern: creative and sales first, engineering deepest, management in between.

Read together, these four case studies turn ordinary job postings into predictive hiring data. A signal that shows up in the numbers before it shows up in a press release.

Next: where AI is actually adding jobs. Stay tuned.