AI Made Cloudflare 100x More Productive. 1,100 People Still Lost Their Jobs.
Cloudflare cut 1,100 jobs on Wednesday. It also told investors that the AI tools replacing those workers have made its remaining engineers up to 100 times more productive — and that its own systems now run on dozens of interconnected AI agents that review code, query internal databases, and route work with almost no human oversight.
The numbers were supposed to tell a clean story. They don't. Cloudflare reported first-quarter revenue of $639.8 million, up 34 percent year over year, with earnings per share of 25 cents beating analyst estimates of 23 cents. Its stock fell 19 percent anyway.
The company is navigating a transition that every infrastructure business will eventually face: the AI tools it sells to customers are the same tools automating its own operations. In November 2025, Cloudflare's internal AI adoption reached what CEO Matthew Prince called an "electric screwdriver" moment — a shift so sudden that productivity gains across teams ranged from two times to 100 times what they were before. According to Cloudflare's own blog post, the company's internal AI usage increased by more than 600 percent in the three months leading up to the announcement. And on the first-quarter earnings call, Prince and his team provided the rest: 97 percent of Cloudflare's engineers use AI coding tools, and every line of code they write is reviewed by autonomous AI agents before it reaches production. Its AI agent workflows run on dozens of servers built around the MCP protocol — a set of rules that lets different AI systems talk to each other and to internal data — and hundreds of centrally managed skill files that define what each agent is allowed to do.
The market's 19 percent after-hours drop reflects competing explanations that analysts have not yet sorted out. Cloudflare's Q2 revenue guidance of $664 million to $665 million came in slightly below the analyst consensus of $665.3 million, per Reuters, which may have driven as much of the selloff as the restructuring announcement itself. Separately, the 600 percent increase in internal AI usage is a figure that comes from Cloudflare, not an independent auditor — it is self-reported, and no independent analyst has yet confirmed the methodology behind it. Bernstein and Morgan Stanley had not published updated notes on Cloudflare as of Wednesday evening. The stock reaction, in other words, may reflect uncertainty about guidance rather than a verdict on the productivity thesis.
The severance terms Cloudflare disclosed in an SEC filing this week are unusually generous — affected US employees receive full pay through the end of 2026 and healthcare through the end of the year — which some analysts read as a signal that the company believes its remaining workforce, superequipped with AI tools, can maintain output without replacing the people who left. CFO Thomas Seifert told investors on the earnings call that Cloudflare's GPU infrastructure now runs at 70 to 80 percent utilization, roughly ten times the single-digit rates typical of hyperscale cloud providers, giving it a cost structure that makes the remaining headcount more productive per dollar spent.
The Goldman Sachs data Cloudflare cited — economists at the bank estimated last year that AI was responsible for 5,000 to 10,000 net job losses per month across the most exposed US industries — positions the company as a first data point in a sector-wide compression, not an outlier. Block, the payments firm led by Jack Dorsey, cut nearly half its workforce in February for similar reasons, according to Reuters. If Cloudflare's model of AI-compressed engineering teams proves sustainable, Namecheap, Fastly, and comparable infrastructure companies face the same calculus, making the 1,100 cuts a leading indicator rather than a one-off.
The company added one million new developers to its platform last quarter, bringing the total to 5.5 million, which Cloudflare points to as evidence that AI is expanding the market for its core product rather than contracting it. What the market is trying to price is the speed of the transition. Prince described the shift internally as going from a manual process to an electric screwdriver — the analogy works because the screwdriver does not eliminate the need for a builder, but it makes one builder do the work of ten. Cloudflare is betting that its remaining engineers, equipped with AI agents that review code, manage infrastructure, and route customer queries, can operate a larger network with a smaller team. The market is not yet convinced that the math works at scale, or that Cloudflare's customers — many of whom are making their own headcount decisions for the same reasons — will maintain or expand their spending on the infrastructure layer above them.
The restructuring will cost $140 million to $150 million in total, with $105 million to $110 million in cash severance, per the SEC filing. The free cash flow guidance for the full year is unchanged. Whether that guidance holds depends on whether the productivity gains Cloudflare described are a one-time compression of the labor curve or the beginning of a permanent restructuring of what a technology company employs versus what it automates. Market participants are broadly watching to see whether the second quarter guidance miss was a blip or the first sign that the same AI adoption Curve that reduced Cloudflare's headcount is also reducing what its customers spend on the network above them.