Cloudflare will cut roughly 20% of its workforce, the internet infrastructure company said on Thursday, framing the reductions as a shift to an "agentic AI-first operating model" rather than a cost-cutting move. The market read it differently. Shares fell 14% in extended trading despite a Q1 earnings beat the same day.
The cuts affect more than 1,100 of Cloudflare's roughly 5,156 staff, the largest reduction in the company's history. In a joint statement, chief executive Matthew Prince and co-founder Michelle Zatlyn said the layoffs reflect a transition to an AI-driven operating model, citing a more than 600% rise in internal AI usage over the past three months.
"Employees across the company from engineering to HR to finance to marketing run thousands of AI agent sessions each day to get their work done," Prince and Zatlyn wrote. "That means we have to be intentional in how we architect our company for the agentic AI era." The pair explicitly disclaimed cost-cutting as the rationale.
What the numbers say
Cloudflare expects to book between $140 million and $150 million in charges in the second quarter, primarily severance and related costs. The Q1 earnings reported the same day were strong on the headline: revenue of $639.8 million beat analyst expectations of $621.9 million, and adjusted earnings per share of 25 cents beat the consensus estimate of 23 cents.
Q2 guidance is where the picture gets messier. Cloudflare projected Q2 revenue between $664 million and $665 million, with the midpoint slightly below the $665.3 million the Street was modelling. Adjusted EPS guidance of 27 cents would be in line with consensus. The combination of a Q1 beat, an in-line to soft Q2 guide, and a one-time charge nearing $150 million is what sent the stock down 14% after hours rather than up.
'Not cost-cutting' is doing a lot of work
The framing matters. A pure productivity-shift narrative is more palatable to investors than a margin-pressure narrative, and a chief executive who can credibly say "we are reorganising for AI, not retrenching" is telling a different growth story than one whose hand was forced. Prince and Zatlyn's joint statement leaned hard into the productivity-shift register.
The market reaction suggests investors were not fully persuaded. A 14% drop on the back of a beat is a significant penalty and one that typically reflects a downgrade to either growth assumptions, margin assumptions, or both. Cutting 20% of headcount, regardless of how it is framed, materially changes the cost base and therefore the margin profile. The agentic-AI productivity story is also, inescapably, a margin story.
It also creates a measurement problem the company will have to live with. If Cloudflare genuinely is transitioning to an AI-operated structure, productivity per remaining employee should rise sharply over the next several quarters, and the headline revenue line should outperform a peer group running historical headcount-to-revenue ratios. If the framing is largely cover for cost discipline, the productivity numbers will look more like the historical norm and the cuts will read as straightforwardly defensive in retrospect. The 14% stock drop is the market's first guess at which version is more likely.
Third in eight days
Cloudflare is now the third significant tech employer in just over a week to announce major layoffs framed around AI:
- Meta on 1 May said it would cut 8,000 jobs to fund its AI infrastructure budget, with chief executive Mark Zuckerberg explicitly framing the trade as compute spending displacing payroll.
- Coinbase on 5 May said it would cut about 700 jobs in an AI-driven restructure, citing AI tools and crypto market volatility together.
- Cloudflare today is the cleanest "AI-only" framing of the three, with the largest stock-market disagreement.
What unites the three announcements is not the underlying cause. Meta's is genuinely about reallocating capital expenditure from payroll to GPUs. Coinbase's coincides with a real crypto trading slump that has historically produced layoffs at the company under any framing. Cloudflare's lands the same day as a Q1 beat, with no obvious revenue distress. What the three have in common is the framing language. AI is now the legible reason of choice when companies of any size cut staff in 2026, and OpenAI's Sam Altman, who described this dynamic in February as "AI washing", is not having his concern aged out by the trend.
The broader 2026 layoff total has now passed 80,000 across the US tech sector. Combined hyperscaler capital expenditure on AI infrastructure is on track to hit roughly $725 billion this year, up about 77% from 2025. The two figures are arriving together, but as Cloudflare's reception illustrates, investors are not necessarily buying the implied causal story when a company cuts a fifth of its staff and tells them it is not about cost.
