Cloudflare announced today it's cutting about 20% of its workforce, roughly 1,100 jobs, as it restructures around AI. That headline alone gets folded into the running tally. 80,000 tech layoffs in Q1 2026, nearly half AI-attributed (per Nikkei Asia's tally). Easy to scroll past at this point.
Don't scroll past this one.
Cloudflare isn't a bloated post-ZIRP hangover company. It's an infrastructure provider with real revenue, real margins, and a CEO who has spoken publicly for nearly two years about how aggressively the company is rebuilding internal workflows around AI. When this company cuts 20%, it's not because the business is broken. It's because the work changed.
"Today's actions are not a cost-cutting exercise or an assessment of individuals' performance; they are about Cloudflare defining how a world-class, high-growth company operates and creates value in the agentic AI era."
— Matthew Prince and Michelle Zatlyn, Co-founders, Cloudflare, in a company blog post
The interesting question is never how many people got cut. It's which ones.
The pattern across the last six months of these announcements is consistent enough now that you can predict it. The cuts hit middle-layer execution roles: second-line support, mid-tier sales ops, manual QA, internal IT operations, content moderation, junior analysts. The roles that, three years ago, were the entry-level pipeline into the rest of the company. Now they're the roles where an AI workflow with a human reviewer ships work faster than a team of ten.
What doesn't get cut: senior engineers building the AI tools, the people running the platform infrastructure, security and compliance, customer-facing roles where the relationship matters, and a small bench of generalists who can move between problems quickly.
If you're a CTO or hiring manager reading this, you're now operating in a workforce model that has three tiers, not four. The first is the senior builders. Engineers, architects, security leads, AI/ML specialists. They're the most expensive and the most retention-critical layer of the company. The second is the generalist operators. Small, senior, cross-functional. They run the AI workflows, supervise outputs, handle edge cases, talk to customers. The third tier is the AI workflows themselves: vendor platforms like ServiceNow, Salesforce Agentforce, Anthropic, OpenAI, plus internal tooling doing the work that used to live in tier 4.
"A lot of the support people that provide support behind them, those roles aren't going to be the roles that, you know, drive companies going forward."
— Matthew Prince, Co-founder & CEO, Cloudflare, on the Q1 2026 earnings call
The fourth tier — the one that used to be the on-ramp — is being compressed out. Cloudflare just made it official for another 1,100 people. Meta and Microsoft together cut roughly 20,000 in April (per Layoffs.fyi). Freshworks did it last week.
This creates an obvious problem nobody wants to talk about. If you compress out the entry-level pipeline today, where do the senior people of 2030 come from? Nowhere good, is the honest answer. Every company is solving for this quarter and assuming someone else will train the bench. They won't. And by the time the senior shortage shows up — and it will, this is the third time I've watched a labor market do this — the firms that bought talent at the price the spot market asks tend to be the ones who survive.
What I'd actually do if I were running engineering or operations at a mid-market company right now:
Stop hiring the role you used to hire. If the job description hasn't materially changed in 18 months, the role probably doesn't exist anymore. Rewrite it before you post it, or you'll spend three months interviewing for a role that AI now does.
Hire one senior generalist instead of three juniors. Until your AI tooling matures enough to actually train juniors on the job, the math doesn't work the other way. Pay for the senior. They'll get more done and they'll architect the workflows the next hire inherits.
Build the bench you'll need in 36 months, now. Even if you only take one junior per year, take the one. Train them inside the AI workflows so they grow up bilingual. Companies that don't do this will be the ones paying $400K for a senior in 2029.
Audit your retention plan for the people who can't be replaced. The senior bench is the actual moat. You should know who, by name, would put you in a hole if they walked tomorrow, and you should be over-investing in keeping them right now. This is the lesson Cloudflare just reinforced.
The story of 2026 isn't going to be "AI took the jobs." It's going to be "AI restructured the org chart, and the companies that read the new chart correctly won." Cloudflare just published their version of it. Read it carefully before you build your own headcount plan for the back half of the year.
The middle is gone. Plan accordingly.