On Wednesday April 22 ServiceNow reported Q1 2026 earnings. The stock slid 14% after hours on an Iran-war-linked subscription revenue miss, and most of the financial-press coverage stopped there. The more interesting sentence came about a half hour into the call, when CEO Bill McDermott told CNBC that ServiceNow expects to start 2027 with the same headcount it started 2026 with — even while continuing to close acquisitions.
That sentence, on a $250B-market-cap SaaS company, is the quietest AI layoff in the sector this week. No severance package is booked. No WARN Act notice fires. Nothing shows up in a Challenger report. But the jobs are gone the same way.
The math McDermott didn’t spell out
ServiceNow ended 2025 with 29,187 employees, per the MacroTrends aggregation of the company’s 10-K filings. That was 2,894 net adds in 2025, or 11% year-over-year growth — roughly the pace the company has kept up for five straight years.
McDermott’s plan is to flat-line that curve. If the historical trajectory would have taken ServiceNow to ~32,400 employees by January 2027, a flat year means roughly 3,200 seats that would have existed, don’t. These are not “cut jobs”; they are non-existent jobs. Economically: the same delta.
What makes it sharper is the “even as we integrate acquisitions” caveat. ServiceNow has closed several acquisitions in the last 18 months (Moveworks, Data.World, and smaller ones). Each brings people. Keeping total headcount flat while absorbing acquired engineers means net attrition in the legacy ServiceNow org is higher than natural. Someone is leaving to make room for the acquired Moveworks team, and not being replaced.
The CEO quote, and what it isn’t
Per the CNBC Enterprise writeup, McDermott framed it as productivity:
“AI will boost productivity so ServiceNow won’t have to backfill open jobs.”
Read that twice. The sentence is doing two things:
- Positioning AI as the reason for the plan. This is the same playbook Microsoft ran on Thursday with its first-ever voluntary buyout, the same one Meta ran yesterday with its 8,000-person May 20 cut. AI is the narrative that makes the headcount decision palatable to analysts and to the remaining workforce.
- Taking layoffs off the table. McDermott gets to say “no layoffs at ServiceNow” on CNBC and be technically right. This is the feature, not the bug, of the attrition-as-policy approach. No severance. No press cycle. No wrongful-termination exposure. No broken glass. Just quieter hallways every quarter.
Benzinga’s characterization — “AI-washing the attrition curve” — is the accurate reading. The AI story makes the headcount plan sound strategic rather than defensive. Whether the AI productivity gains actually materialize on the inside is a separate question that shows up in 18 months of engineering velocity data, not in a Q1 press release.
How this differs from Microsoft and Meta this week
Three Fortune 500 companies ran three variants of the same playbook inside 48 hours:
- Meta (April 23): 8,000 involuntary cuts, 16 weeks severance, 6,000 open roles left unfilled, effective May 20. The loud version. Headlines, WARN notices, SEC filings.
- Microsoft (April 23): voluntary buyout for ~8,750 people under the “Rule of 70” (age + tenure ≥ 70). Booked in fiscal Q4. First in the company’s 51-year history. The dignified version.
- ServiceNow (April 22): no cuts, no buyouts, no severance. Just don’t backfill. The stealth version.
The economics end up in the same place: fewer people, same revenue target, higher operating margin. The political surface area is radically different. ServiceNow’s version is the one that leaves no fingerprints. TheStreet noted that McDermott also delivered a blunt warning to new grads on the same call — the entry-level door is the one most affected by “no backfill” policy, because entry-level roles are the ones most often opened and re-opened on a high cadence.
Why the stealth version scales faster
If you are a CFO at another SaaS company watching the three variants, ServiceNow’s is the one you copy first. The attractiveness stacks:
- No cash cost. Buyouts are real dollars. Severance is real dollars. Natural attrition is 0 dollars.
- No news cycle. “Company X laid off Y workers” is a headline. “Company X didn’t post Z new reqs” is not a story.
- No retention risk. Buyouts create bimodal attrition — your best people take them, your weakest people don’t. Attrition-only leaves you with whoever was already staying. The signal is flatter.
- Reversible. If the AI productivity thesis fails, you can quietly resume hiring with zero optics damage. Microsoft’s 51-year-streak break is very public. ServiceNow’s backfill freeze is invisible.
Expect the shape to spread. Workday, Atlassian, Datadog, Snowflake, HubSpot — the mid-cap SaaS cohort that just watched ServiceNow test the language on a live earnings call — will have the same framing in their own May and June calls. Some will use the word “AI”; all will talk about “productivity” and “efficiency.” The thing being described is the same thing.
Why LostJobs cares
- Not-hiring and firing are economically identical for the person who wasn’t hired. For a 22-year-old with a fresh CS degree applying to ServiceNow in the summer of 2026, the ~3,200 seats that were supposed to exist and don’t are the single largest data point in their job search. No press release will name them; the Indeed and LinkedIn req counts just go down. This is the labor-market transmission mechanism that is hardest to see and hardest to legislate.
- Attrition-as-policy shifts the burden to the insider. If headcount is flat but the work isn’t, the remaining employees absorb the delta. “AI productivity” has to literally cover 3,200 people’s worth of output, or the existing 29,187 run hot to make up the difference. In enterprise-software history, the latter is the more common outcome.
- The Q1 2026 trio is a template. Meta, Microsoft, ServiceNow. Loud, dignified, stealth. Every Big Tech and Big SaaS company now has three reference implementations for “how to pay for AI out of payroll.” The stealth one scales best because it has the lowest friction. That means more copies of it, coming faster.
McDermott’s line — “we don’t have to backfill” — is the 2026 version of “we’re pausing new hires.” The pause is doing the same work the cuts used to.