Cloudflare's CEO Explained Exactly Which Jobs He Replaced With AI. Most Owners Should Read It, Then Do Less.
Builders, sellers and measurers: a growing company cut 20% of staff and named the roles. A small-firm reading of the op-ed, and the risk it skips.
Most companies that cut jobs "because of AI" are vague about it. In May 2026 Cloudflare's CEO, Matthew Prince, wasn't. In a Wall Street Journal op-ed titled "How I Choose Which Cloudflare Employees to Replace With AI", he named the roles.
What happened
Cloudflare laid off more than 1,100 employees, close to 20% of its workforce, in May 2026, People Matters reported, citing a company memo. Prince wrote that it didn't happen because the company was struggling: Cloudflare was posting record revenue growth and strong free cash flow. A company memo cited in coverage said internal use of AI tools had risen more than sixfold in three months.
Builders, sellers, measurers
Borrowing from Peter Drucker, Prince sorted roles into three groups:
- Builders create the product.
- Sellers win and keep customers.
- Measurers count, check and coordinate: internal audit, finance, legal, compliance, operations, middle management, revenue recognition.
"The vast majority of those we laid off last week were measurers," he wrote. His argument: AI can now monitor and analyze an organization continuously, so fewer people are needed to do it. Cloudflare cut layers of middle management, merged operations into one group, significantly reduced marketing and automated parts of finance. He said the company was closing its books faster and auditing risks continuously instead of periodically.
He also wrote that the company had a record number of open positions and would keep hiring builders and sellers: "If an engineer on my team can now be 10 times as productive, I'm going to hire as many as I can find."
What's useful for a small firm
The sorting exercise is worth doing. List every role and the hours each spends building, selling and measuring. In a small firm, measuring is often scattered: the owner reconciling, a manager compiling reports, someone chasing timesheets. Those tasks are good AI candidates because they're repetitive and checkable.
Redeploy before you remove. In a ten-person firm, the "measurer" usually does three other things too. Automating the reporting may free them to sell or serve clients, which is the growth Prince says he's investing in.
The risk the op-ed skips
Measuring isn't only counting. Audit, compliance and finance exist partly to catch mistakes and problems in the work of the very people they check. If AI does the measuring, someone independent still has to own the result and be able to question it. In a small firm, that is usually the owner or the outside accountant. Don't let the check and the thing being checked end up as the same system run by the same person.
And Cloudflare is a fast-growing public company with deep engineering resources. Meta, with even more, scaled back its own plan (our write-up), and Ford rehired veteran engineers after automation fell short (our write-up).
A small-firm version
- Map hours by builder / seller / measurer for each role.
- Pick one measuring task, such as month-end reporting or reconciliations, and run AI on it with a named reviewer.
- Measure time and errors for 60 days.
- Move freed hours to client or sales work. Revisit roles only after that.
Use our AI ROI calculator for step three and the AI governance checklist to keep checks independent.
A role-sorting worksheet for owners
Do not label a whole employee a builder, seller or measurer. Most small-firm jobs contain all three. Break the role into recurring tasks and estimate the hours in each bucket for a normal month. Then ask a different question for each task: can the output be checked quickly, and what happens if it is wrong?
- Building: making the product, delivering the service, solving the customer's actual problem.
- Selling: earning trust, discovering needs, following up and keeping relationships alive.
- Measuring: compiling, reconciling, reporting, routing, reminding and checking whether rules were followed.
Start with measuring tasks that are frequent, structured and reversible. Keep a named person accountable for reviewing samples and investigating exceptions, especially where the measure affects pay, taxes, safety, legal duties or a customer's money. Independence matters: the person rewarded for a good number should not be the only person designing the system that produces and certifies it.
If automation frees five hours, assign those hours before the trial starts. “More selling” is too vague; two prospect calls, same-day lead follow-up or a backlog of client reviews is a use you can see. Otherwise the saved time is absorbed invisibly and the experiment proves nothing.
The "should I replace staff with AI?" series: Meta's plan stalls · Ford rehires veterans · Cloudflare's "measurers"
Frequently asked questions
Which employees did Cloudflare replace with AI?
CEO Matthew Prince wrote that most of those laid off were 'measurers': roles such as internal audit, finance, legal, compliance, operations, middle management and revenue recognition.
How many people did Cloudflare lay off?
More than 1,100 employees, about 20% of its workforce, in May 2026, while reporting record revenue growth.
Should small businesses copy Cloudflare?
Copy the thinking, not the cuts. Sort roles by what they do, test AI on measuring tasks with a human still accountable, and remember that checks exist to catch the people being checked.
