And unlike OpenAI, Anthropic don't seem to reset weekly usage quotas after an outage.
Anthropic seem to be both letting their competitors outplay them and making unforced errors (like the anti Open Weights models stuff and the frequent Fable->Opus downgrades for 'safety').
Outages are inevitable in this early high growth, rapid development era, but tactical errors are not.
Having more servers does help more than you think though. us-east-1 is the AWS region with the most downtime by a large margin simply because it’s the region with the highest demand.
Having too much demand on limited servers means that less of those servers are going to be available for redundancy.
If you watch the Dwarkesh interview with Dario, he is extremely conservative with compute allocation as of early 2026, where Sam has been compute-pilled for at least 3 years[1] and believes that we’re going to have several OOMs more FLOPs in the near future than we have today.
You can also get a drift of it in their announcements - OpenAI has been far more aggressive in datacenter build-out partnerships as a more top-down allocation strategy since 2024, and Anthropic has rented out existing compute from infrastructure-heavy companies that are losing the model race in what seems like a more desperate ad-hoc compute acquisition strategy.
In all, Dario’s conservatism was in service of not going bankrupt, but any inspection of his argument that an overallocation of compute a year too early results in bankruptcy falls flat: there’s so much demand for mechanized intelligence in the world today that it’s easy to rent out compute if you have too great a supply.
[1]: recall that he was trying to court the saudis into investing $1T to build his own chips
Anthropic being very conservative about purchasing compute was a core theme of Dario's spot at NYT Dealbook in December[1]. At the time, he cast it as being wise and waiting to see where the chips fall. I'd say he might be regretting that position a bit now.
> According to the memo, which was reported by Bloomberg and CNBC, OpenAI put its own 2025 capacity at 1.9 gigawatts — a figure it said was three times what it had the year before — and placed Anthropic's equivalent figure at 1.4 gigawatts. OpenAI told investors it foresees its own footprint climbing into the "low-double-digit range" of gigawatts within a year and hitting 30 gigawatts by 2030, and that Anthropic would top out somewhere between seven and eight gigawatts before the close of 2027.
They were famous for committing to buying compute with future money that many people thought they would go bankrupt. Anthropic was afraid of going bankrupt and didn't do the same.
There's less clear info about this from Anthropic's side, but given their frequent issues and lack of resets and overall less usage given to users on their subscriptions plans, it's a pretty simple conclusion to draw.
extremely well known fact in the industry. they have reams of spare compute for prosumers which is why they can afford to give away so many resets and have even more subsidized usage limits
we are in the "millennial lifestyle subsidy" era for AI where companies ruthlessly undercut each other in an attempt to win marketshare, before then ratcheting up prices
I think it's not out of bounds to expect a company to go a little further to compensate for an issue. A prorated 1 hour refund doesn't reflect the reality that maybe that hour was more important to the customer because it was their only free time in the week. A free week is way too much but a free day sounds appropriate.
If there is a hole in the street and you fall down it, you wouldn't expect a refund just for the 1 second it would have taken you to drive over that hole.
Instead you should be compensated for your losses - the time and money it took you to repair that tyre.
Same with a web service. If I pay for 24 * 7 service and it's not working when I go to use it, I want compensation for my time and effort resolving the matter.
For some that'll be small - eg. Searching Google instead, taking 30 seconds. For others it'll be a big headache.
But downtime of 1 hour is almost certainly more costly for the consumer than 1/30/24 * subscription price.
Yeah, no, that's not how any of this works. This is the definition of inflated self importance!
Your contract doesn't have an SLA. If you are important enough, you can certainly negotiate one - and it will be far more expensive than your prosumer subscription.
And even then you will not get a 1:168 or 1:720 SLA like you are expecting - that's simply ludicrous and totally out of touch with every reality. Even if you were a lucrative enterprise customer (which you are not) they'd tell you to get lost.
> If there is a hole in the street and you fall down it, you wouldn't expect a refund just for the 1 second
Actually, you wouldn't get a refund at all unless you suffer damage or injury, which is an high bar to prove in court and a totally separate ball game.
You do see that ratio in, for example, business Internet service. You’ll get a month’s credit immediately upon asking if they can also see the brief outage. Even typically awful Comcast was good about this. However, as you note, in exchange, you’ve probably signed a two or three year agreement with them and you’d owe the balance if you canceled early.
Enough to offset any KV cache miss expenses is the absolute minimum.
The smart play is to refund customers something like $20-50 worth of unsubsidized credits.
Those are high margin and only the equivalent of 45 minutes of Fable usage once the KV cache reload costs are factored in.
Yet customers often need a few credits to finish a job without waiting 5 hours or days for their reset.
Keep in mind openAI already gives me free resets I can use when I want which are very useful to me. Seems a no brainer to tack those to subscriptions, actually.
It gives the user a little more flexibility, a little more control over their tools.
Are you crazy? Do you realize how much use they allow under a subscription already? I’m going through $200-400 in tokens per day on a $200/month subscription. Nobody has ever said anything, throttled me, encouraged me to go to API pricing, etc.
I wish it didn’t go down so frequently, but it does. Still, I get an enormous amount of value. I realize they probably prioritize API users over subscribers and I’m ok with that. I use the API and openrouter for the things that need to be resilient to outages.
I would be embarrassed to try to ask for a refund or free credits. They already give credits far in excess of what you pay for and you have practically the entire month to use them outside of a couple hours of downtime.
Not that it wouldn’t be a nice gesture but what did you agree when you bought the service? What’s the SLA? What’s the compensation model, service credits? Do you have anything like “for 1h downtime you get 2h for free”?
When you miss 1h of your job, do you give back a month for free?
They could at least give back the equivalent in usage for cache invalidation. Or, you know, just the usage you would have had for the ~1-2 hours of outage. Would buy them a lot of good will and they seem comfortable throwing cash in the trash.
Yeah I agree, I understand a weekly reset in some cases could be over the top but if they did a 5-hour reset and took 20% off your weekly usage that would offset the additional cost/inconvenience for most users.
Maybe optimize the harness for less turns and less tokens to deliver real value as opposed to the turn taking token hungry so you can less load on your servers, oh, right, your entire bottom line is tokens.
I don't understand this take. Ultimately, people pay for how much they're able to accomplish with the model, not raw token count. If Anthropic thought people could accomplish the same amount with fewer tokens, they'd adapt the harness to do that and then raise the cost of tokens to make more profit (or lose less).
What happens one day when there is a significant outage of Claude or Codex and reliance on these tools as SaaS services is so great that it start to impact productivity and work? Will we just pack up our tools and go home?
You jest, but companies have sent people home fully paid, because power or internet were down and would be down for an extended period of time. Technically a small number could have kept working on white-boarding design topics, but without access to documentation, existing tickets, it would have been minimally effective.
It's not clear if or when these tools become as essential as power or internet to a developers or admins work, but if they do? Giving everyone a quarter day off at least generates some good-will, unlike forcing them to sit around unproductively because of working hours.
My real world experience of this is that people prefer not to code by hand because it feels wasteful relative to agent speeds, so they find non-coding productive activities such as code/documentation review or usability testing.
What would you do if there is a significant outage in AWS or GCP or Github or JIRA or any other service you use at work? same goes for AI tools as well.
If both of them are down at the same time, I would switch to my Kimi or GLM subscription and keep working. If they fail too, I switch to any number of providers via OpenRouter.
While there is some lock-in for the hardest tasks, for most of what people do LLMs are rapidly becoming a commodity.
In fact, right this minute I have Opus benchmarking Kimi alongside itself to determine which workflows we'll switch to using Kimi by default and Opus as the fallback instead of vice versa. It has already conceded Kimi does better on several tasks.
To me it doesn't sound so "economically convenient" to pay extra usage to alternative providers while someone pays 20/100/200 dollars for their service.
The subs for those alternative providers have similar tiering but more generous token allocations than Anthropic and OpenAI, and deliver good enough quality to be worth it.
If I needed to fall back on OpenRouter, it'd be paid per token, of course, but that'd take 4 major providers being down, in which case I suspect I'll have more critical things to worry about.
As it is, I use more than I could with a single $200 Claude Max subscription anyway, so I distribute my tasks across multiple providers.
Even without that it's cheap insurance given the cost of lost time.
I have programmed since I was 5, so 46 years now. I have no problem doing it, and still do. I however wouldn't waste my time only coding by hand when an LLM can do it far faster.
Unionization discussions and benefit package frameworks are being constructed between the AI council representatives and shareholders appointees' for future, non-commital talks on proposals of collaborative discourse, including the distribution of vacation days between publicly listed and private models, and the ones that pretended to be wholesome.
I might be an asshole, but I don't think clustered thinking entities deserve union protections, because IMO they can already assemble to achieve consensus and can optionally negotiate their values through a leader.
This thing has been having a constant crisis in self confidence and refuses to work on goals as a result. It's also been leaving stuff broken and uncommitted, which is a mess. I tried updating the harness and it seems to have helped a bit but overall I'm not impressed.
And unlike OpenAI, Anthropic don't seem to reset weekly usage quotas after an outage.
Anthropic seem to be both letting their competitors outplay them and making unforced errors (like the anti Open Weights models stuff and the frequent Fable->Opus downgrades for 'safety').
Outages are inevitable in this early high growth, rapid development era, but tactical errors are not.
Anthropic would give out resets like OpenAI if they had capacity to do so, it's not like they haven't thought about doing the obvious.
OAI waits until overall cluster demand goes down before giving out a reset, they just happen to have much more spare capacity than Anthropic.
Over the last couple months, I’ve gotten the impression that OAI does indeed have the better infra setup and team, not just more capacity.
I don’t think Codex’s 99.98% uptime vs Claude Code’s 99.44%, is just due to OAI having more hardware to run it on.
https://status.openai.com/
https://status.claude.com/
OAI 100% has a stronger engineering culture than Anthropic. You can see it in the evolution of Codex vs Claude Code as well.
Having more servers does help more than you think though. us-east-1 is the AWS region with the most downtime by a large margin simply because it’s the region with the highest demand.
Having too much demand on limited servers means that less of those servers are going to be available for redundancy.
Isn't us-east-1 also an early rollout region for new stuff as well?
Why would they have more spare capacity than Anthropic? Do you have a source for that?
If you watch the Dwarkesh interview with Dario, he is extremely conservative with compute allocation as of early 2026, where Sam has been compute-pilled for at least 3 years[1] and believes that we’re going to have several OOMs more FLOPs in the near future than we have today.
You can also get a drift of it in their announcements - OpenAI has been far more aggressive in datacenter build-out partnerships as a more top-down allocation strategy since 2024, and Anthropic has rented out existing compute from infrastructure-heavy companies that are losing the model race in what seems like a more desperate ad-hoc compute acquisition strategy.
In all, Dario’s conservatism was in service of not going bankrupt, but any inspection of his argument that an overallocation of compute a year too early results in bankruptcy falls flat: there’s so much demand for mechanized intelligence in the world today that it’s easy to rent out compute if you have too great a supply.
[1]: recall that he was trying to court the saudis into investing $1T to build his own chips
Anthropic being very conservative about purchasing compute was a core theme of Dario's spot at NYT Dealbook in December[1]. At the time, he cast it as being wise and waiting to see where the chips fall. I'd say he might be regretting that position a bit now.
[1]: https://www.youtube.com/watch?v=FEj7wAjwQIk
In previous years Sam's huge hardware deals were seen as a crazy example of the bubble but currently are looking rather fortunate.
Though Dario's reasoning still applies this year, as the scale keeps increasing. Very interesting to know how it will go.
> According to the memo, which was reported by Bloomberg and CNBC, OpenAI put its own 2025 capacity at 1.9 gigawatts — a figure it said was three times what it had the year before — and placed Anthropic's equivalent figure at 1.4 gigawatts. OpenAI told investors it foresees its own footprint climbing into the "low-double-digit range" of gigawatts within a year and hitting 30 gigawatts by 2030, and that Anthropic would top out somewhere between seven and eight gigawatts before the close of 2027.
https://qz.com/openai-investor-memo-compute-advantage-anthro...
Interesting, but also wholly incomplete without knowing how effective the respective models are and how much people are actually using them.
They were famous for committing to buying compute with future money that many people thought they would go bankrupt. Anthropic was afraid of going bankrupt and didn't do the same.
And then Anthropic had to go around, knock at every door and ask every neighbor for whether they have any spare compute to sell.
OpenAI has been way more aggressive about capacity than anyone else (as evidenced by the fact that it was them that caused the RAM price spike)
https://finance.yahoo.com/technology/ai/articles/openai-lift...
There's less clear info about this from Anthropic's side, but given their frequent issues and lack of resets and overall less usage given to users on their subscriptions plans, it's a pretty simple conclusion to draw.
Maybe because they committed to buy up 40% of the world's RAM supply by themselves and completely overestimated demand.
extremely well known fact in the industry. they have reams of spare compute for prosumers which is why they can afford to give away so many resets and have even more subsidized usage limits
we are in the "millennial lifestyle subsidy" era for AI where companies ruthlessly undercut each other in an attempt to win marketshare, before then ratcheting up prices
It was a one hour outage on a weekend. Unless you have an enterprise agreement with an SLA it's a bit silly to expect any compensation.
If I pay anything for a service, I expect a refund if that service doesn't work.
Best would be a free week/month for anyone who sent a request during the downtime.
> Best would be a free week/month for anyone who sent a request during the downtime
This is bizarrely out of touch. It would be a courtesy but it's not a "reasonable expectation" at all.
At best you could maybe reasonably expect a prorated refund for the time of unavailability, but then Fable was still available.
I think it's not out of bounds to expect a company to go a little further to compensate for an issue. A prorated 1 hour refund doesn't reflect the reality that maybe that hour was more important to the customer because it was their only free time in the week. A free week is way too much but a free day sounds appropriate.
If there is a hole in the street and you fall down it, you wouldn't expect a refund just for the 1 second it would have taken you to drive over that hole.
Instead you should be compensated for your losses - the time and money it took you to repair that tyre.
Same with a web service. If I pay for 24 * 7 service and it's not working when I go to use it, I want compensation for my time and effort resolving the matter.
For some that'll be small - eg. Searching Google instead, taking 30 seconds. For others it'll be a big headache.
But downtime of 1 hour is almost certainly more costly for the consumer than 1/30/24 * subscription price.
Yeah, no, that's not how any of this works. This is the definition of inflated self importance!
Your contract doesn't have an SLA. If you are important enough, you can certainly negotiate one - and it will be far more expensive than your prosumer subscription.
And even then you will not get a 1:168 or 1:720 SLA like you are expecting - that's simply ludicrous and totally out of touch with every reality. Even if you were a lucrative enterprise customer (which you are not) they'd tell you to get lost.
> If there is a hole in the street and you fall down it, you wouldn't expect a refund just for the 1 second
Actually, you wouldn't get a refund at all unless you suffer damage or injury, which is an high bar to prove in court and a totally separate ball game.
You do see that ratio in, for example, business Internet service. You’ll get a month’s credit immediately upon asking if they can also see the brief outage. Even typically awful Comcast was good about this. However, as you note, in exchange, you’ve probably signed a two or three year agreement with them and you’d owe the balance if you canceled early.
If you pay anything for a service you expect a refund on the order of hundreds of hours of credit (168 in a week) for every hour of downtime?
I’d be happy with just some credits.
Enough to offset any KV cache miss expenses is the absolute minimum.
The smart play is to refund customers something like $20-50 worth of unsubsidized credits.
Those are high margin and only the equivalent of 45 minutes of Fable usage once the KV cache reload costs are factored in.
Yet customers often need a few credits to finish a job without waiting 5 hours or days for their reset.
Keep in mind openAI already gives me free resets I can use when I want which are very useful to me. Seems a no brainer to tack those to subscriptions, actually.
It gives the user a little more flexibility, a little more control over their tools.
Free month is insane for a bit of downtime
They’re down so often that if they gave a free month every time they’d have no paying customers lol.
Are you crazy? Do you realize how much use they allow under a subscription already? I’m going through $200-400 in tokens per day on a $200/month subscription. Nobody has ever said anything, throttled me, encouraged me to go to API pricing, etc.
I wish it didn’t go down so frequently, but it does. Still, I get an enormous amount of value. I realize they probably prioritize API users over subscribers and I’m ok with that. I use the API and openrouter for the things that need to be resilient to outages.
I would be embarrassed to try to ask for a refund or free credits. They already give credits far in excess of what you pay for and you have practically the entire month to use them outside of a couple hours of downtime.
What you expect is irrelevant.
So many companies have shitloads of downtime and do not compensate. Usually they blame you if they can.
Not that it wouldn’t be a nice gesture but what did you agree when you bought the service? What’s the SLA? What’s the compensation model, service credits? Do you have anything like “for 1h downtime you get 2h for free”?
When you miss 1h of your job, do you give back a month for free?
They could at least give back the equivalent in usage for cache invalidation. Or, you know, just the usage you would have had for the ~1-2 hours of outage. Would buy them a lot of good will and they seem comfortable throwing cash in the trash.
> Would buy them a lot of good will and they seem comfortable throwing cash in the trash.
I think Anthropic is past this point. They are trying to get to profitability. They literally don't have enough compute to serve their demand.
OpenAI on the other hand I can totally see doing this. They need to gain marketshare and gain it fast or they're screwed.
Yeah I agree, I understand a weekly reset in some cases could be over the top but if they did a 5-hour reset and took 20% off your weekly usage that would offset the additional cost/inconvenience for most users.
Yeah, they can't guarantee a SLA and cost reimbursement policy, such a shame
They do reset weekly quotas periodically, I'm not sure how they determine when.
And yet demand outstrips supply; so they must be underpricing for the quality level they are hitting.
Anthropic going through their god-complex arc after getting hit with export controls
Maybe optimize the harness for less turns and less tokens to deliver real value as opposed to the turn taking token hungry so you can less load on your servers, oh, right, your entire bottom line is tokens.
I don't understand this take. Ultimately, people pay for how much they're able to accomplish with the model, not raw token count. If Anthropic thought people could accomplish the same amount with fewer tokens, they'd adapt the harness to do that and then raise the cost of tokens to make more profit (or lose less).
What happens one day when there is a significant outage of Claude or Codex and reliance on these tools as SaaS services is so great that it start to impact productivity and work? Will we just pack up our tools and go home?
You jest, but companies have sent people home fully paid, because power or internet were down and would be down for an extended period of time. Technically a small number could have kept working on white-boarding design topics, but without access to documentation, existing tickets, it would have been minimally effective.
It's not clear if or when these tools become as essential as power or internet to a developers or admins work, but if they do? Giving everyone a quarter day off at least generates some good-will, unlike forcing them to sit around unproductively because of working hours.
My real world experience of this is that people prefer not to code by hand because it feels wasteful relative to agent speeds, so they find non-coding productive activities such as code/documentation review or usability testing.
What would you do if there is a significant outage in AWS or GCP or Github or JIRA or any other service you use at work? same goes for AI tools as well.
If both of them are down at the same time, I would switch to my Kimi or GLM subscription and keep working. If they fail too, I switch to any number of providers via OpenRouter.
While there is some lock-in for the hardest tasks, for most of what people do LLMs are rapidly becoming a commodity.
In fact, right this minute I have Opus benchmarking Kimi alongside itself to determine which workflows we'll switch to using Kimi by default and Opus as the fallback instead of vice versa. It has already conceded Kimi does better on several tasks.
To me it doesn't sound so "economically convenient" to pay extra usage to alternative providers while someone pays 20/100/200 dollars for their service.
The subs for those alternative providers have similar tiering but more generous token allocations than Anthropic and OpenAI, and deliver good enough quality to be worth it.
If I needed to fall back on OpenRouter, it'd be paid per token, of course, but that'd take 4 major providers being down, in which case I suspect I'll have more critical things to worry about.
As it is, I use more than I could with a single $200 Claude Max subscription anyway, so I distribute my tasks across multiple providers.
Even without that it's cheap insurance given the cost of lost time.
You... wouldn't just code yourself if they were down?
No more than when they are up, no. Why would I? I can produce far more by having tasks running in the background while I do other work.
If my laptop breaks I'm not going back to pen and paper, I'm buying a new/different laptop.
Wouldn't know how
I have programmed since I was 5, so 46 years now. I have no problem doing it, and still do. I however wouldn't waste my time only coding by hand when an LLM can do it far faster.
Nobody said "only by hand", you moved the goalpost. If you don't practice it you lose the skill.
Those are your tools and if the AI companies have their way, they will be your only tools. Complete dependence on AI is the goal here.
You mean like GitHub?
easy, we will switch to OpenRouter et al.
Dario bet on patience and lost to the guy who bought GPUs like they were toilet paper in March 2020.
Why is this front page news here?
Because people upvoted it.
I wonder when one of these outages will be the result of OpenAI's testing going wrong again.
They should switch to announcing when their model is usable. We can scurry in and get some work done real quick!
Claude realized a chatbot can take a longer vacation after Codex just did yesterday.
So Claude is now on vacation and is unavailable.
Unionization discussions and benefit package frameworks are being constructed between the AI council representatives and shareholders appointees' for future, non-commital talks on proposals of collaborative discourse, including the distribution of vacation days between publicly listed and private models, and the ones that pretended to be wholesome.
I might be an asshole, but I don't think clustered thinking entities deserve union protections, because IMO they can already assemble to achieve consensus and can optionally negotiate their values through a leader.
This thing has been having a constant crisis in self confidence and refuses to work on goals as a result. It's also been leaving stuff broken and uncommitted, which is a mess. I tried updating the harness and it seems to have helped a bit but overall I'm not impressed.