Probably true. But that’s not the position we are in now. Apple is much better aligned with users interests than any one else, at least for short/medium term.
Oh, me, me! I spent a few years being responsible for a significant bit of DMA review and CYA and responses to regulators.
I’ve read all of it, multiple times, and been grilled by EU regulators (vicariously, via corporate lawyers).
It still boils down to general guidelines that it’s impossible to know if you’re violating before the fact, and they will not even approve/reject proposals in advance. It’s basically “go read the act yourself, and ship what you think is compliant, and you’ll know whether we interpret the words the same way by whether or not we fine you.”
> It still boils down to general guidelines that it’s impossible to know if you’re violating before the fact, and they will not even approve/reject proposals in advance. It’s basically “go read the act yourself, and ship what you think is compliant, and you’ll know whether we interpret the words the same way by whether or not we fine you.”
Companies want to know exactly where the line is so they can figure out how to comply with the letter of the law while doing as much as possible to get around the spirit of the law. This has been demonstrated over and over again. It isn't the job of the regulator to help companies with this process.
So you’d be cool with speed limit signs that said “hey, don’t go too fast” and no specific limits? And the cops decide who to pull over on vibes, reputation, mood?
I’m more of a rule of law person myself. If there’s a law that must not be broken, and breaking it results in penalties, it seems insane to me to not specify it in advance.
Sure, big tech is largely evil. Arrest ‘em, find them, IDGAF.
But pretending that DMA and related regulations provide enough information to ensure compliance is willfully ignorant. The regulations are designed to allow selective enforcemen.
>speed limit signs that said “hey, don’t go too fast”
Yeah we already have that. We have the words 'SLOW' on the road that ask you to slow down from the current limit for the hazard ahead, but pulling you up on this is officers discretion.
Bad example, as there is a posted speed limit above which you are positively breaking the law. Discretion may lower the limit.
The DMA doesn’t have the objective measures. It’s all discretion, all subjective, all post hoc.
Which, cool, some people like the idea that police target those people and need flexibility to make life harder for undesirables in ways they would never do to high status people. I don’t personally like that, and I don’t think tech regulation should work that way.
Hi Simon, nice article. The parent there may be making the same assumption I am, that large enterprise _never_ pays sticker price.
Also, to just color in the picture here, as I haven't seen it mentioned elsewhere, there is a very large Saas company at the moment who has given everyone unlimited tokens on Claude. And they have a dashboard showing who spends the most. So the "budget" went from about USD500 per per person (split between Claude and cursor) in Jan to... Well a soft limit of USD100k... Per month... Per person.
People can still see the top line sticker price on their spend, but honestly I can't believe that the Saas is paying that full price when the invoice comes in.
That said, there are some finance reports which are probably dropping soon where we will find out!
> The parent there may be making the same assumption I am, that large enterprise _never_ pays sticker price.
I shared that assumption until yesterday, when I found out that it wasn't holding for LLM pricing from OpenAI and Anthropic. That's what inspired me to write this piece.
I think those token leaderboards are an obviously terrible idea and will go extinct very quickly now that people are paying attention to costs.
But the feature list at https://claude.com/pricing#team-&-enterprise literally lists "tiered incentives on committed spend" and "non-standard terms" as perks of the sales-assisted Enterprise plan. Maybe "non-standard terms" could mean "we dance for you if you pay", but what would "tiered incentives on committed spend" mean besides "we can negotiate on price if you bring the volume"
Volume discounts may be available for high-volume users. These are negotiated on a case-by-case basis.
* Standard tiers use the pricing shown in Model pricing
* Enterprise customers can contact sales for custom pricing
And there are discounts available through "Claude Platform on AWS":
Anthropic rates your token usage in USD at standard per-model, per-feature rates, applies any negotiated discount, converts the result to CCUs at $0.01 per CCU, and reports the CCU quantity to AWS Marketplace hourly. Your AWS bill shows a single CCU line item.
> With the pricing change, customers of Claude Enterprise, a two-year-old bundle of products meant for large companies that now includes Claude Code and its work assistant, Claude Cowork, will have to pay for the amount of computing capacity they consume while using the software on top of a monthly flat fee of $20 per user, an Anthropic spokesperson confirmed.
The sentence you quoted doesn't say anything about the token price for enterprises matching the sticker price. It just says that enterprises are paying a consumption based price (presumably billed as some $/token), on top of a per-seat fee.
> I found out that it wasn't holding for LLM pricing
You're correct. When a type of cloud service grows large enough and has a few competitive suppliers, enterprise pricing tends to coalesce with the large buyers paying around the same price for the same thing. While that might be lower than the publicly cited rate card, the private price similarly large customers pay ends up being similar.
One reason is that the very largest, long-term enterprise customers are so valuable, they can command MFN clauses ensuring no one else is paying substantially less for the same thing, then the rest of the rate card for smaller customers flows down from that. There's a strong disincentive for vendors to cheat or allow big disparities between similar classes of customers because the number of people involved on both sides of these deals is large enough that word will get around eventually.
Large scale enterprise sales and purchasing in a given sector tends to be rather circular. Account execs move to other vendors and call on the same customers, while purchasing agents can move to other customer firms. Personal relationships, reputation and credibility matter. Lying to screw a large customer over just to make one commission or quarterly quota can be a very bad long-term career move. Sometimes purchasing agents or executives quietly compare notes off-the-record with their peers from other similarly-sized firms. After dinner drinks at industry association meetings and trade shows can be quite productive in terms of verbally exchanging 'market insight' with peers.
When there are significant pricing differences, its usually due to different volume commitments, SLA/QoS guarantees, payment terms and other material factors which justify the difference. Source: been there, done that inside a top ten valley tech company. Once was in a meeting where a newly minted EVP tried to get a long-time senior account exec to pressure a huge customer by being semi-dishonest. The account exec schooled the EVP on the fact that the EVP could only make him unemployed for about 8 hours but that huge customer not wanting to work with him could make him unemployed forever. :-)
I do know of moderate-size companies deploying OSS LLMs on their own GPU clusters, for ownership/security/maybe cost reasons. I'm somewhat surprised F500 companies are apparently just handing over all their data to the model providers.
Could be fantastic for small shops while it lasts. The big guys have to pay 10x for precious tokens.
And "large" just means that AWS will assign an account manager to talk with you. I was at a start-up who spent $300k/year on AWS and that was enough to get special attention and discounts. Enterprise pricing is confusing.
Claude is so in demand at the moment that there aren't really volume discounts. Anthropic sets the terms and you either accept them or get lost they have that much of a lead (mindshare/desirability wise).
They pay sticker price. There may be exceptions for very very large companies like Amazon or Microsoft which have their own deals where they rent out compute in return for usage.
I don't know exactly I never worked there and I was IT not design/qa/factory floor. My understanding was slow, non innovative, not efficient, and prone to complain. The ex-Boeing people we had were great though. I think it's probably just normal 150k employee companies end up toxic stuff. Just know we had a quota.
I think it really depends on the motivations of the business. Some are more R&D and innovation driven. Getting "buy-in" is technically necessary, but trivially easy as long as the biggest cost is only development time. If it's a bad idea, it eventually fades away as other priorities take over.
There is no one singular "corporate environment". This is especially true when a lot of people working there tend to not job hop much. Time both grows that particular work culture, and keeps those people ignorant.
Right but Google has loads of market leading services they use to deliver those ads, and Facebook has a crappy dated website and dumb ideas that lose them money which they brand the company on.
Do you know you’re just talking to an LLM? Everyone else in this post also seem oblivious to it or maybe they just don’t care? Why do I even read comments anymore sigh
Just forget everything you've learned about it previously, use 6.2+, non isolated by default and every upcoming feature turned on.
And your intuition will be spot on.
DRBD is more of a live sync, and it's great stuff, as long as you set it up BEFORE you need it, and you need it frequently. If you want to keep a second copy of your data on another system, up to the second(ish), it's a great choice.
If, however, you just want a copy of a block device on another system, like for weekly backup (our case), it's probably overkill. Especially as to keep it truly consistent you need to run in the mode where writes are acked only once the remote AND local devices have it.
My VMs are running on ganeti, which has a mode where the backing device can be DRBD and written to another host. Which works great if you have the extra disc space and can deal with the latency. Also allows you to live migrate VMs between the two hosts.
In my case I ultimately want the copy off-site, so DRBD isn't really a great fit.
DRBD is very good stuff though, I've used it for decades for HA database servers and the like.
reply