Forgot your password?
typodupeerror
Businesses AI

Cloudflare To Cut About 20% Workforce As AI Adoption Reshapes Operations 20

Cloudflare plans to cut about 20% of its workforce, or more than 1,100 employees, as it restructures around an "agentic AI-first operating model." Reuters reports: Cloudflare CEO Matthew Prince and co-founder Michelle Zatlyn said in a message to employees that the company was reimagining every team and function to operate in what they described as an agentic AI era. Cloudflare said the job cuts reflect a redesign of internal processes and roles, rather than a response to employee performance or short-term cost pressures. The company added that its own use of AI has increased more than sixfold over the past three months, prompting major changes in how teams operate.
This discussion has been archived. No new comments can be posted.

Cloudflare To Cut About 20% Workforce As AI Adoption Reshapes Operations

Comments Filter:
  • by Somervillain ( 4719341 ) on Friday May 08, 2026 @11:36AM (#66134108)
    No, like everyone else, Cloudflare simply overhired in the past and forsees reduced revenue opportunity in the future...probably due to industrial trends as well as macroeconomic pressures, like tariffs, increased energy costs, wars, etc. Or they simply overhired in the past. However, they are choosing to commit fraud to investors, like most companies and link it to AI instead of their diminished future prospects and past management choices (which I wouldn't call mismanagement...better to hire for potential opportunity and correct when you're mistaken...but investors may disagree)

    Does AI reduce labor need? If you use it, you'd know that's not the case. Claude isn't cause a 20% efficiency gain across an organization. If your programmers didn't suck, they would barely notice. The majority of a decent programmer's time is spent finding edge cases, negotiating requirements, validation, etc. My time writing Java is about 5% of my day...because I know how to do it well and can do it very quickly. The rest of the time, I am meeting with stakeholders, validating my work, finding edge cases, troubleshooting error reports, etc.

    The "how?" is the easiest question, the only one an AI can help with. The "what?" (am I doing) and "why?" are the huge costs.

    However, let's say I'm wrong...Claude is causing a greater than 20% reduction in time needed. Why isn't CloudFlare using their 10x AI-fueled developers to crush their massive competitors? Akamai and Fastly are huge....So cloudflare is happy with their current marketshare and wants to maintain it for 20% lower cost?....vs crushing all competition? I call bullshit.
    • by Anonymous Coward

      No, like everyone else, Cloudflare simply overhired in the past

      Standard operating procedure:

      1. Hire thousands of people you don't actually need. This creates the illusion that your company is successful and growing. PROFIT!

      2. When profits start going down, get rid of all those people you never should have hired in the first place. This creates the illusion that you are efficient and cutting costs. PROFIT!

      3. CEO gets a big raise.

      4. Lather, Rinse, Repeat.

    • Miss forecast? Announce layoffs and hopefully the stock recovers. That's a CFO playbook. I don't think they've been building their agent-centric architecture for long enough to know how many engineers are needed. I suspect they just needed a bump in the numbers, and let that weight fall on workers as usual.
      Edge networks/CDN/firewall providers seem to be companies where they have dozens of engineers on call 24/7 to put out fires and apply fixes for tickets created by hundreds of thousands of clients. Automat

      • Edge networks/CDN/firewall providers seem to be companies where they have dozens of engineers on call 24/7 to put out fires and apply fixes for tickets created by hundreds of thousands of clients. Automating some of that or all of that means less engineers are needed to manage resources, firewall rules etc. I don't know if they have a core dev team working on traffic pattern analysis/bot detection/network and infra tools, but I'm guessing all those engineers are safe.

        You're incorrect on that part. All those tools? They don't require AI. They can be done algorithmically. If it can be automated, it has been before LLMs were publicly available. I have worked for teams automating these precise operations. There's too much money on the line to take a human out of the loop. There are various ML automations...perhaps some LLM ones...but LLMs are slow and these 24/7 techs are there to react in seconds, not minutes. However, the state of the art is that these automations

  • by rsilvergun ( 571051 ) on Friday May 08, 2026 @11:40AM (#66134116)
    And we are just not allowed to talk about it because 90% of the media is owned by billionaires and they got huge huge tax cuts and tons of deregulation from the current administration and the current administration wants to keep up the illusion.

    What's weird is how many people are participating in the illusion. The administration polls at around 40% overall approval rating despite everything going on and despite the fact that individually the policies are extremely unpopular.

    Eventually reality comes calling but if they can hold it together until after the midterms once you cast your vote that's it you don't get any take backs. The American system makes it basically impossible to undo the damage from a stupid vote until the next election which for a senator takes 6 years.

    And again all you have to do is confuse people for about 2 to 4 weeks leading up to an election and you're all set you can keep screwing them over.

    I don't know how you solve that. We need to be teaching critical thinking in schools but parents don't like that because they don't like it when Johnny comes home and starts asking questions about their religion or political beliefs or whatever stupid ideas they have in their head. People want their kids to be smart enough to get a job but not so smart they can no longer relate to them or agree with them. So they purposely stunt their growth. Like how in that book brave New world the people destined to simple do repetitive labor were given chemicals in the womb to limit their intelligence...
  • CloudFlare was an aggressive global internet surveillance and privacy invasion operation. Now it's an AI-powered aggressive global internet surveillance and privacy invasion operation.

    Why don't I feel excited about it?

  • Just like square this is more AI washed separation. I have yet to see a single AI model that truly saves time. The recent case of Oracle eliminating jobs is related to the fact their Special Purpose Vehicle (to move debt slightly off book) is growing to a dangerous point. Right now they have BILLIONS in debt, that can only be serviced if OpenAI can pay them more than 70 Billion on a regular basis. This is insane.

    Who honestly believes these tools are actually valuable. I've seen people using Claude and ot
  • by crmarvin42 ( 652893 ) on Friday May 08, 2026 @12:58PM (#66134306)
    I wonder what they will do when the cost of AI increases?

    We all know that AI companies are selling their services at a loss. Often on a cost-of-compute- basis, but even more so when you factor in model training costs incurred with investor cash. And that is even before we account for how the shortages of relevant hardware and server space for running all of this are driving up the costs of memory, chips, etc. Or the fact that the energy crisis is only getting started, and will impact literally every part of the value chain for addressing the current and future demand.

    Most of the sunk costs to date, have been funded with investor cash, but those investors are going to start wanting to get paid back with a strong multiple of their investments to date. That means, as companies reorganize around the use of AI - at the current prices - they are creating a potential nightmare of cost forecasting and control when the AI vendors all decide it is time to start generating that pay-back by sticking the screws to their customers. This is CLASSIC ENSHITIFICATION.
    • by ranton ( 36917 )

      I wonder what they will do when the cost of AI increases? We all know that AI companies are selling their services at a loss. Often on a cost-of-compute- basis, but even more so when you factor in model training costs incurred with investor cash. And that is even before we account for how the shortages of relevant hardware and server space for running all of this are driving up the costs of memory, chips, etc. Or the fact that the energy crisis is only getting started, and will impact literally every part of the value chain for addressing the current and future demand.

      In 1998, 1 Mbps of bandwidth cost $1200 per month. Today it is about 10 cents. The past never perfectly predicts the future, but I wouldn't be on the side who thinks AI won't be significantly cheaper in the near future. My Claude subscription costs be $200 per month today and gives me $2500-3000 worth of tokens per month. But in 5 years that same amount of usage will probably be a few hundred, and in ten years it will probably be $50.

      • That makes no sense at all. If the company is spending $2-3k, for something they charge only $200, then it is an introductory price. To get you hooked. They have to charge you $3-4k at some point of they want to make money. Their suppliers want to get paid, and their investors want to stop losing money eventually. By ask means take advantage while the deal is good, but donâ(TM)t be so naive as to expect the deal to be good forever.
      • Compute costs are not improving by leaps and bounds like they were in the 90's. Today, most companies struggle to get 5% improvement per year. AI is still in the early days, but it's still very much hardware bound and it's unlikely to mature at an exponential rate.

        Besides, the AI market has gone bonkers, and hundreds of billions are being poured into infrastructure. Investors want to make all that money back.

        I don't suspect subscription costs to be going down any time soon.

      • In 1998, 1 Mbps of bandwidth cost $1200 per month. Today it is about 10 cents.

        I feel your argument lacks merit. In 1998, the average size of an image file on the web was between 2k to 12k, and the average web page was around 100K. Today the average web page is 3M.

        1Mbps is certainly cheaper today, but it is also certainly less useful.

"I think trash is the most important manifestation of culture we have in my lifetime." - Johnny Legend

Working...