Claude.ai down(status.claude.com) |
Claude.ai down(status.claude.com) |
Maybe Mythos went in too hard and removed the physical network too?
downtime is always going to 'scale' poorly against loads that require a lot of hardware thrown at them, even with lots of good fail-over -- probably worse for the small vendors because they don't have the contracts supplying them with hardware first so availability is already at a premium for them.
so, I guess i'm saying yeah I hope frontier-level-models get out soon in the open arenas, but I suspect the same or similar level of exclusivity will exist as long as they take that much compute to operate.
There is no way to know whether it is a good or bad business decision just because they can go down when a third party goes down. For example, if you save $50 million a year by firing half your employees and replacing them with AI, but you lose $10 million a year because your site goes down when Claude goes down, then you made a great business decision.
I honestly don't care nearly as much as I used to, because I used to be more upset over this. Now, I simply wait to see how much is enough to rile up average Joe and Jenna.
Mostly I think its that management does not blame the person who picks AWS. Its another iteration of "no one got fired for buying IBM/Microsoft".
It is also an issue at other levels: if all a county's businesses rely on AWS (let alone its government) then that gives the US huge leverage over you (sanctions would shut down your economy).
Most of humanity doesn't think this, nor I doubt any devs like the current state of affairs where 4 companies dictate the direction of technology in this country.
“Web 2.0” was supposed to be a web of interoperable applications with features exposed by APIs, where you could assemble little pieces of functionality into new and novel configurations. It was cool for a little while, but it was never sustainable. There was an enclosure of the commons (they were never really the commons), and now we’re all digital serfs.
It was a beautiful dream, while it lasted!
It is a weird thing to respect come to think of it... Having an accurate status page should be the baseline, but that's the world we live in.
One of the least transparent companies I've been a customer of.
And I’ve experienced 500 error codes in Claude code that lasted more than thirty minutes but less than an hour that never showed up on the status page as an outage
Not a lot of them provide uptime in % values, but Anthropic doesn't either.
I’m thinking for example of Apple having a multiple hour outages a couple weeks ago, preventing anyone be from installing debug apps on devices. Yet, everything was green the entire time.
To another's point, yeah, it's sad that Claude taking a brief hiatus just halts workflow. I imagine my whole team just started packing for a beach trip. Then again, I suppose if the CPA office computers go down, everyone doesn't dust off giant ledgers and quill pens and start working manually.
If you create a plan it follows it closely
yall hilarious
neeeeeeeeeeeeeeeeeeeeeeeeed
They’ll need Claude to fix it during an outage, but Claude is down too!
https://arstechnica.com/ai/2026/03/after-outages-amazon-to-m...
There are just some things you should not vibe code your way out of.
He just needs to finish his lunch in 30 mins before going back to work.
I use Anthropic for 90% of my GenAI needs, software development, but I have never used Claude.ai for anything other than auth.
I took the time to stand up and do some stretching. Good use of time.
Anthropic: 9.
World: 9 nines? That's awesome!
Anthropic: No, a singular 9. 90-something percent uptime.
Replicating the basic functionality for most SaaS is trivial, it's the "everything else" part that you're actually paying for.
And you thought GitHub or us-east-1 outages were bad.
They have a very unusual workload in that they need to write to disk a lot more than what is typical, making them harder to host for a generic cloud provider. And also Azure sucks so bad it's almost impressive GitHub has any uptime at all!
Isn't there a thought here that you don't host your status page on your infra itself, and if for example they are using hetzner etc. and hosting their own vibe coded one, then they might need an engineer so might as well have Saas whose whole job is this.
I could've maybe recommended one of those status software[0] which actually runs on github actions but github themselves seem to have some downtime :p
[0]: https://github.com/upptime/upptime (for anyone interested)
If a team is so dependent on Claude that they’re paralyzed when it goes down, I’d worry that they’re in over their heads. Opus 4.6 is amazing but still has limits. You need to know what you’re looking at and how to send it in the right direction, as well as when to reject its output.
I'm doing the work of an entire team now. I can still do the work of one person by hand, but that's not acceptable anymore.
And if my IDE or compiler (or computer!) stopped working because it requires a connection to the mothership I'd be livid. But I guess the cloud-everything, subscription-everything model has successfully made people accept an objectively worse world.
"It's weird how when internal cloud goes down, no one can function until it's back."
Any dependency is like this, it's not the first, it won't be the last.
I am pretty sure most people deep in AI tool use various LLM providers anyway.
Your system is down as often as AWS. When you are down your lost sales are shared between 10 of them. When AWS is down you get all their lost sales.
Obviously very simplified, but you get the point. There might be a huge gain in being up when others are down.
> Betting your business on AWS being flakier than whatever alternative provider you use is probably not a good idea.
You are not betting your business on it. You are betting the consequences of downtime only.
AWS does not seem to be all that high reliability out of the box. You can use multiple availability zones etc. but you can do the equivalent elsewhere.
> On the infrastructure side: OpenAI signed non-binding letters of intent with Samsung and SK Hynix for up to 900,000 DRAM wafers per month, roughly 40% of global output. These were of course non-binding. Micron, reading the demand signal, shut down its 29-year-old Crucial consumer memory brand to redirect all capacity toward AI customers. Then Stargate Texas was cancelled, OpenAI and Oracle couldn’t agree terms, and the demand that had justified Micron’s entire strategic pivot simply vanished. Micron’s stock crashed.
[0] https://adlrocha.substack.com/p/adlrocha-how-the-ai-loser-ma...
Most companies aren’t going to opt for a bespoke solution to things if something tried and true already exists. Maybe the really small simple applications will be affected, but the ones with hundreds of features and years of experience will be fine.
True, All as in everything. I think Anthropic is extremely silly for thinking that and maybe even a bit delusional.
But to take the devil sides here, maybe some might be a better accurate term as to be honest, Status pages are a mission critical. Imagine this page landed 404 and the panic then vs now. Also I can be wrong, I usually am but status page innovation is more devops oriented than software itself. Basically how do you keep the status page website as having the best status time of itself (imagine it goes down) and its needed 24x7 as outage can happen anytime.
I do feel like status pages are more the exception than the norm in how mission critical they can be, again I can be wrong. Also you can self host status page easily yourself if you wish (UptimeKuma is awesome!) on a 7$/yr vps and there are so many free options available out there and its very much a competitive space to be in and basically no lock-in for the most part.
The steelman version of "SaaS is dead due to AI" isn't that SaaS companies will disappear, it's that competition will greatly intensify, to the extent that it becomes a commodity business with thin margins, rather than the money printer it is today.
I don’t think the house factory analogy makes sense for multiple reasons. I subscribe to multiple LLM providers and switch between them all day. I could sign up for a dozen more to provide GLM 5.1 if I wanted to as well. I can even run lesser models locally on my machine.
This is nothing like a single factory because I can switch to a new provider in minutes with a credit card.
Which doesn't mean the open models will definitely succeed, it just means they have more of a shot than the open social networks ever did
I don't know about that. More usage means more support, which means more docs and open source projects, wrappers, harnesses built around them etc.
Way less demand to build tooling around open weight models if they remain hobbyist.
1. Using AI helps as part of the training process.
2. All the prompts going to openai/claude is a gold mine.