Just another form of vendor lock-in. If your business model is mostly/entirely dependent on an external party, that should be a well understood risk.
The only people winning are selling shovels
Dude, it’s 2026. We don’t sell shovels, we sell shovel subscriptions.
Tiered shovel subscriptions.
Dynamically priced
Oh absolutely, Anthropic has dynamic token usage during peak hours.
Or rather the right to use shovels under ToS that can be changed on a whim.
And you can sell at least one shovel to somebody digging themselves into a hole
I am responsible for gathering information on AI to determine whether we should use it for our next project. The ask was to use it for a critical process task. Immediately in my head I was like “no, we are not using AI at all”, but I obviously need quantifiable data. This is just another thing to add to my list of why using AI for core processes is one of the stupidest things you could ever do.
Well you could also spec out a few machines for local LLMs as a sensitive alternative to show the higher ups. “This is what it’ll cost us if we don’t want to be caught with nothing but our dicks in our hands when a vendor decides to shut us down for actually using the shit we’re paying for” and “it’ll end up saving us money eventually” (if you can male the case for that, you’ll have to do your own calculations).
Keep in mind the top of the line models will require some 600-700 GB of VRAM IIRC, may want to check ollama for examples. And you’d want redundancy of course, not a single machine.
Capex will usually seem more sensible to businesses than opex since it’s a one time thing, but this should be big enough to deter them unless you work at a really big company.
But also, what type of task is it? Perhaps AI is not a bad fit, just LLMs. There are plenty of decent use cases for other types of AI. For an example you could tell if something is a hot dog or not with pretty good accuracy.
Oh no! How did this happen? …I mean, how exactly did this happen? Is there a tutorial on how other engineers at other companies can replicate this?
Just so they can avoid the same mistakes of course. Engineers hate mistakes.
Aaaaaand example #99999… Of why tech sovereignty is so important. The moment you start outsourcing your control, you become vulnerable to this exact kind of action by a company.
Everybody got sucked into the cloud “magic” for years, but now we are seeing the monster emerge more and more as proprietary technology enshitifies.
Luckily, there is a boom happening across the FOSS world, more and more people are finally waking up to the principles of software freedom and actual ownership.
May it continue to grow, as the corpos struggle and wither.
I was working as a sadmin (like a sysadmin but more alcoholism) when the cloud butt became all the rage.
Suddenly nobody wanted to host services on the hypervisor down the road, administered by someone you could throttle call in a crisis. Nobody wanted to hire a monkey to keep their local tubes clean and run the basic stuff they needed.
Everyone could tell you that once they had your overbuilt shit locked in to their very specific apis and services they had you by the short and curlies and by god were they gonna squeeze for all you were worth.
Alas, nobody cared because initial offerings were cheap and your stupid magento storefront had to be webscale.
Now 6 companies control the internet and everything else is going that way too.
Everyone could tell you that once they had your overbuilt shit locked in to their very specific apis and services they had you by the short and curlies and by god were they gonna squeeze for all you were worth.
On-prem solutions don’t necessarily protect companies from this either though. Anyone staring down the barrel of a Broadcom renewal for on-prem VMware licenses knows this pain.
Broadcom’s screwing over of VMware has been the biggest accelerator of migration into the cloud in the last 5 years.
There are FOSS hypervisors that are more than adequate for almost everyone’s useage. I would not advise anyone to make any single company a critical part of their infrastructure unless you are tightly integrated in a mutually beneficial arrangement.
If you have your own sysadmin then you don’t tend to get as fucked, alternatively migrating hypervisor software is a fuckload easier than migrating from a cloud service provider.
There are FOSS hypervisors that are more than adequate for almost everyone’s useage.
While I wish that were true I didn’t find that to be the case.
For home use? Aboslutely. For Small businesses, probably, but labor costs rise noticeably in maintaining those alternate FOSS hypervisors. That can be a dealbreaker for lots of companies which swings the pendulum back to cloud (or Microsoft on-prem hypervisor).
When the Broadcom/VMware apocalypse occurred I looked at all other hypervisor options both FOSS and commercial and found none that were close to VMware’s feature offerings for large enterprises. The best for most orgs would be HyperV only because of existing MS licensing the orgs had would cover most of the new license burden.
We used to use KVM and qemu. There was no serious overhead maintaining them.
How many VMs were you running? How many regions and what level of geographic redundancy were you offering your org? Were you serving any type of organization that had regulatory compliance/audit requirements (FDA, HIPAA, PCI, DoD, SOX, etc)?
Idk thousands? we were a hosting provider lol. Don’t want to dox myself. Not sure how regions come into it, I mean if you can write shell and some orchistration language you’re golden for anything.
We had some PCI stuff, I relapsed smoking because of getting through it haha. We were also halfway through getting the Australian government PII/gov contract thing when I left.
Most people suck at passing audit compliance because they try to box tick rather than explain how their tailored systems meet and exceed the requirements.
That’s what happens when you are renting your very skills from a company. You’ll hone nothing and you’ll be happy.
but but ai better, ai future, we pay moni to all companiea Nd buy ai or we will be left without any growth - pleaz buy all ai- ai goof for making woled better place because it makes billionaires richer and they will definitely use that fo donate for charity
( blinking twice Elon musk and Mark Zuckerberg told me to say that, I’m being held at gunpoint)
Good twist on that one.
Or… taps mic… don’t fucking rely on AI for your business! Play stupid games, win stupid prizes.
This has nothing to do with AI.
Don’t rely on software or workflows or really anything that you can’t easily switch if said company decides to stop doing business with you.
If you do, it better be a strategic partnership where something like this can’t happen.
In this case, their workflows should have been AI provider agnostic or had a way to continue functioning if Claude went down.
This definitely has to do with AI. Because CEOs are losing their stupid minds over it. I agree with you in principle, but let’s not lose sight of the fact that this specific technology is what CEOs are drooling over. Even in my company I had to tell the owner/CEO, “What problem are you trying to solve with AI?” His response was his mouth being open with a dumb look on his face.
So no business should rely on AI (or, to your point, any software) that it becomes detrimental to their business or workforce should that access be revoked.
Yes, this has everything to do with AI, because this is an AI vendor locking out a customer from their ordinary workflow.
At the same time, this is a generalizable example not limited to AI, where any form of vendor lock-in on a critical business function becomes a potential point of failure when the vendor drops the customer or stops working. It’s true of a cloud provider, an email provider, an ISP, any software provider that can revoke access/authority, or even non-tech vendors like a landlord or a temp agency or an electric utility.
Landlord lock in, that’s a funny one. So what’s the plan, move to another town because you can’t find a 4 story building available for next week to avoid rent increase?
This is actually a pretty common concern for businesses on dealing with whether and how to protect themselves when installing improvements, business-critical equipment, or other hard-to-move stuff on land or in a building without a long term lease in place.
The tenant deals with it by either building out a portable infrastructure to where they can move their business quickly if need be, or by protecting themselves legally to where the landlord can’t kick them out on a short notice, by negotiating a long term lease.
Vendor lock in for today’s software is almost impossible to avoid unless you are running on owned bare metal which is not really an option for many mid size companies.
It can be hard, but its not impossible for many things.
Like if you use AWS S3 there are S3 compatible APIs at cloudflare and likely other cloud providers.
If youre using a service that offers cloud functions and one offers the programming language you want to use, but others dont, maybe its better to use the more common language that all the platforms offer even if its not youre preferred choice.
If you were using Slack, have a plan to switch quickly to Teams if something goes wrong and slack drops you so you can get comms up quicker.
For those where alternatives arent an option, it should be a very conscious choice with the knowledge it might bite you in the ass with no quick recovery.
I absolutely agree. Many just don’t think the benefit of being nimble is worth it. Glad to see it being a bigger discussion.
I was (un?)fortunate to work at a company early on in my development career that ran into a problem where poor design choices (not mine) limited our ability to be nimble. Ive been able to carry that lesson on. Not that im perfect at it either though haha.
It has worked out to my benefit many times though.
We’re in a period where the tools, agentic systems in this case, are gated by large companies.
This is like if IBM or Cray in the 60s through 90s only allowed rental of mainframes that they owned, and they can cut you off.
That wasn’t the case then, but just like Google shutting down the father’s entire Google account cause the pediatric doctor wanted a photo of the kid who had a rash to see if they needed to be brought into the ER or a cream, then got his phone (Google Fi), email (Gmail), and all his paperwork backups (drive) cut off… When you don’t own the infrastructure you live at the whims of things you can not even appeal to.
This is a story about people and companies putting their entire business workflows in the hands of big tech who really don’t care about anyone.
So, AI drama aside, the moment your life or business is fully dependent on an unreliable partner, this is what happens.
Many commenters were quick to point out that he should never have coupled his company so closely with Claude to begin with, a reasonable critique by itself. However, it’s worth noting that the story could have easily been the same if it had instead been Amazon Web Services, Azure, or an authentication provider like Okta.
You are so close, you almost got it!
This is true for any company using 3rd party services. I worked for one that used a 3rd party messaging service to send out mfa texts to users. The company was hacked and went offline, so we couldnt send any mfa codes… and of course, they had no plan b.
In business, always have a backup
This is true for any company using 3rd party services.
It’s like when a streamer / content-creator gets “deplatformed” from whatever service and they had put all their eggs in that basket.
Yeah that’s what tends to happen when monopolies and oligopolies form. That streamer should be making money equally from a diversified pool of YouTube, patreon, Spotify and discord and at least 3 others. If they started earning more in one platform they should pull back to keep the income balanced
/S
60 employees were dead in the water, as reportedly their daily workflows rely on the AI assistant’s
Is that a joke? 60 employees do not know how to do their job? This is not Anthropic’s problem.
I throw any bullshit task into AI. I’m to produce a monthly report on my strategic wins and goals for the next month. I throw it in AI, don’t read it, paste it in the Google doc, send it to the PM who sends it to my boss who also doesn’t read it (or uses AI to read it).
Now I know how to write it but writing this report would take me a day or two if I carefully did it, or 3 mins with AI.
But you are not “dead in the water” if you suddenly can’t use AI to do it this month, right?
my point was, AI dying would affect my productivity but not because I don’t know how to do my job.
I get that, but the first comment didn’t say they don’t know how to do their jobs because their productivity was reduced, it was because they were “dead in the water”. I read that as “unable to do their jobs without AI, period”.
Also it’s a bit funny because what’d actually reduce your productivity wouldn’t be AI dying, it would be a useless bullshit task your boss gave you although he doesn’t even read the report, according to you.
On the contrary it’s exactly what Anthropic wants.
Not necessarily, they could have just a piece of the entire build/deploy process that requires some access to Claude to complete and have no real easy way to turn it off. Like multiple CI/CD steps reaching out for validation or something stupid.
something stupid.
Say no more.
CI/CD If right should be including a
try: AI except: no AINow that’s good stuff
This is the nightmare scenario for any team that built their whole workflow around a cloud API. No warning, no clear reason, no real support path. just a Google form and 60 people sitting on their hands.
The uncomfortable truth is that “terms of service” at this scale is just “we can pull the rug whenever.” Anthropic isn’t unique here either. OpenAI, Google, all of them have the same opaque enforcement problem. It’s a big part of why I’ve been building tools that run on local inference by default. Not because cloud is bad, but because your users shouldn’t be one vague policy complaint away from a complete outage.
Local gives you continuity even when the upstream disappears.
Either they didn’t pay, they found an exploit, or, more likely, someone at Claude was reviewing their conversations. Take note, any business that cares about IP or confidentiality.
I’ll bring two theories to the table.
a) they got caught distilling for their own models b) they re-sold their $200/mo plans as APIs
Just continue coding using the natural neural networks in the brains of those 60 employees until the problem has been resolved and/or another AI provider selected. It’s not like Claude invented coding. Sure, it’s a pretty useful tool. But it is possible to research obscure APIs and develop software manually.
Shut up. Nobody wants that. Work psh
Isn’t it hilarious how capitalists are trying to force us all into literal ‘nobody wants to work anymore’ territory and we’re not even onboard
Because they want the “efficiency” of firing shitloads of people, but not the onus of actually paying their taxes so those people can have UBI.
Ironically, this is a great case study to illustrate the value of Chinese models. They’ve released a number that are on par with Claude’s latest models under “open weight” licenses that would allow you to run them yourselves if you wanted to, or to hire some other third party to provide API access. It wouldn’t matter what the original company’s “usage policy” is in that case.
There are a couple of Western open models that aren’t bad either, but they tend to be aimed at a smaller and simpler use case than Claude.
What models exactly? And what kind of hardware do you need to run them? Also, are there any GitHub repos that replicate Claude projects?
The one currently making the headlines is Kimi K2.6, on the benchmarks it’s just short of Opus 4.7. It’s a trillion-parameter model so it won’t run on desktop computers, but it’s something a company could run on reasonably buildable servers for their own use.
For local use, I’ve been finding Qwen3.6’s 35B parameter model to be uncannily good. Gemma4 is also good, that’s one of the Western ones. These models won’t do the sort of heavy lifting that Opus can do but you don’t need that heavy lifting for all tasks.
They are not as capable as opus, and that sadly matters.
Kimi K2.6 is close to Opus. It beats Opus 4.6 on the benchmarks, so if Opus 4.6 was sufficient for your needs then Kimi K2.6 should be on par.
If you literally can’t access Opus because Anthropic cut you off I suspect that matters more than a slight difference in benchmarks.
Quick, another fix of the LLMs! Let’s not think about what the downtime means for the industry.
https://bannedbyanthropic.com/
I believe the word is capricious. Everything cloud based is at the whim of someone else.
There are ways to mitigate against that, but ultimately if it’s not yours…it’s not yours.
You’re going to see a lot more of this and other forms of fuckery as the VC money dries up.
https://www.wheresyoured.at/four-horsemen-of-the-aipocalypse/
Yup. when the purveyors have to finally charge for what it costs, these fanboys will flee quickly
Now this company can see which employee can actually still program, and which is just a “AI Prompt Engineer”.
