Hacker Newsnew | past | comments | ask | show | jobs | submit | darkwizard42's commentslogin

Seeing a lot of tricks similar to how ridesharing companies tried to be "profitable" before going to IPO. Caveat: Thing have materially improved but really Uber is carried by its insane Ads margins

The idea of removing model training from your costs is a little wild tbh.

The profitability of being able to serve a query wasn't really under question (nor is the margin expected to be anything less than 80%+) I think.


> The idea of removing model training from your costs is a little wild tbh.

Yeah, I didn't believe they'd claim something like that. But yes indeed, from the article:

> Anthropic's gross margins are above 80% before accounting for revenue shared with distribution partners, including Amazon (AMZN.O), opens new tab, and the cost of training its model,

Is this how all AI companies calculate if they're profitable or not, by removing the highest costs? What a circus.


> Anthropic has told shareholders that its adjusted operating income will be positive for a second straight quarter, the Financial Times reported on Sunday, citing multiple people with knowledge of the matter.

Note this claim is about „operating profit“, which commonly is the revenue - operating expenses (COGS, rent, payroll). This does not include RnD cost.

>Anthropic's gross margins are above 80% before accounting for revenue shared with distribution partners, including Amazon (AMZN.O), and the cost of training its model, the newspaper said.

Gross margin is typically (revenue - COGS) / revenue. Thus, both statements above seem generally in line with commonly accepted accounting standards.


If you are sharing revenue (royalties, licenses based on revenue, costs that scale directly with revenue) doesnt that count as cost of goods sold?

Typically R&D is not part COGS. It’s absolutely part of the bottom line when you will typically recognize the cost over some period of time to try to get a true picture of the business.

It’s also easier to strip it out of the picture to think about how much it costs to serve the next token. If you can have great economics to serve the next token (profitable) you can always figure out ways to further reduce your R&D costs.

Now they are absolutely intertwined but I don’t think this is ever as big of an issue that people make it out to be. Replace token with any widget, this is how businesses measure themselves.


This is how all companies calculate gross margins, which is why they're called gross. Gross profit is revenue minus the cost of goods sold. For example, if you sell wheat for $1 and spend $0.20 of labor and raw materials (e.g. fertilizer) to grow it, your gross margin is 80%. This will not include equipment like tractors. AI companies, like all others, include this in their quarterly reports because they are legally, professionally, and customarily required to.

The real circus is commenters on HackerNews thinking this number means Anthropic is cooking the books or that it's actually profitable. Gross margin is meaningless for an AI company, since most of their expenses are R&D and infrastructure (the two things excluded from gross margin), but they have to report it anyway.


Andy Fastow would be envious.

I believe this is a new accounting method called “phantasmagorical accounting.”

The way I read it, they are convincing the investors that they can fool the larger population convincingly. At the end of the day the investor term is misnomer for big institutional investors, given that these people are managing other people money where they always make out a certain percentage of fees despite the outcome.

> The profitability of being able to serve a query wasn't really under question

In one sense, yes, but I do see people question it regularly.


I still question it. It's very light on details by just casually dropping they have 80% margins. I smell bullshit on this.

It is certainly not above them to play accounting tricks to pretend to be anywhere near profitable.

If you create a machine that can turn a dollar into 5, you don't dillute ownership of the machine, you use your fabulous profits to expand production. Anthropic, on the other hand, raises money like crazy, and seems desperate to IPO.


I think part of the big push to "slow down AI development" is to add some sort of regulatory pressure that will give them sort cover to train less models and slow their burn rates

I can’t count the number of times I’ve heard variants of ‘they’re losing money on every query’ and ‘I’m getting 10k worth of tokens for $200’ over the last year. People clearly believed serving margins were -ve

I think the latter half of your post is missing

Hehe, he reached limit of his tokens /s

> The profitability of being able to serve a query wasn't really under question (nor is the margin expected to be anything less than 80%+) I think.

HN had long debates about whether AI inference could even be affordable from a compute perspective.


> idea of removing model training from your costs is a little wild tbh

It's one of several metrics and tries to estimate steady-state profitability. It's the only one being leaked because it's the most sensational one. But don't assume cash-flow profitability is negative just because you don't know it.


I didn't realize how much money Uber makes from ads.... Why does every business devolve into an ad platform?

Might be because most hit their maximum growth but need to keep growing indefinitely or risk becoming a “mature” company?

Because consumers don't care much about ads and prefer them to even minor cost differences.

Any business where people are looking at a screen should probably sell ads.

Vomit. We should ban 90% of ads. They are mostly net negative on society.

Well then you haven't listened to Ed Zitron or any of the other AI bubble doomers. His contention is that its worthless and they lose money on every query.

But isn't he taking all the costs into account, that created the experience? Rather than just literally the inference/serving infrastructure? Bananas way of calculating things if so, doesn't match reality at all.

isn't that when the training costs are actually accounted for?

Not for operating profit. Training is a one-time cost that gets amortized across every query.

And either way, the training of new base models will eventually slow from the current frantic pace.


>Not for operating profit. Training is a one-time cost that gets amortized across every query.

But people are saying its not part of the calculation to generate the cost of a query...

>And either way, the training of new base models will eventually slow from the current frantic pace.

Sure, maybe.


Zitron is worthless–lying about numbers and not correcting the record when you're called out means you aren't trustworthy. Worse than that if you directionally agree with him, which I do.

He makes his money and fame by spewing bulshit as many grifters before him.

Which unfortunately probably hides the real truth. That large labs do have potential problems with long term profitability.


Who do you think is being tricked exactly? I can guarantee it isn't the team of lawyers and analysts at their investors firms.

I think the evolution of the harness and ability to preserve loop context outside the context window has made running these kinds of agentic experiences easier.

sorry so many buzzwords to say, the capabilities to do this kind of work are more accessible and easier to manage, so now it works!

Good to see, and agree they were severely overhyping their product back then.


The average person is not using a calendar and is not getting more than 1 relevant personal email per week.

Most people use their personal email extremely sparingly and it is inundated with promotional / account-related materials which are already handled rather deftly by most mainstream email providers


Ha, tell me you don't have children without...

The average kid in an average suburban public school is not getting this amount of digital communication. If they are, they still aren't using a Google Calendar and are managing the complexity in much simpler ways (paper calendars)

I think they mean the parents. I got 5 emails from the kids school today, and have regular calendar events related to the kids. So yeah this is for parents

To be fair I do get fucking hammered by school emails, and not just school school but dance school.

But like, an AI would just tell me what’s in the email and ask me what I want to do about it which is almost always ignore it and I know what’s in the email because I have eyes.


lol no one is getting personal emails, maybe some people get one every few months, but most probably never (not taking into account other than order/shipping confirmations, login codes, etc)

I can only imagine you have no kids, serve on no boards, do not volunteer, and play no sports.

The average person is not serving on a board and is definitely not volunteering.

Sports communications at most travel/rec leagues is via group chat / Whatsapp / FB group.


First of all, the average parent is definitely doing some volunteering, at least in the US. From rec sports to scouts to church stuff to local committees, it's very rare for someone to be totally checked-out on all those counts.

And as a bit of anectata: I play two sports and my child plays multiple. Every single one of those groups or teams relies on email, even when other modes of communication are layered on.


I don't see you publishing something which further reduces the memory footprint?

Given the project is open source, have at it! See if you can help improve it instead of complaining. This is a significant improvement and a cool project!


I run actual Winamp for my music streaming, and windows task manager claims it's using 4.1 MB.

Even that seems pretty high, but whatever, but given y'all are spending hundreds of mbs, I think I'm ok. Yes, spotify has features my subscription music service doesn't have... but the one I use lets me stream with Winamp.


I'm not complaining and it is a cool project (from what I can tell). But I think it's fair to say 200 MB memory usage is a lot without providing source code as a proof. Just by feel when you recall what were computers capable of when they did not have 100s of MB to spare.


I remember when I was scrounging memory sticks to get to 512 mb for my windows xp setup. And I was able to play Unreal: Return to Na Pali[0] on that.

[0] https://unreal.fandom.com/wiki/Unreal_Mission_Pack:_Return_t...


The problem is that it requires vast sums of money to influence politics.

That is the antithesis of how politics should function (people are the interest, not moneyed companies)


Some payment is withheld unless competitors (Youtube, TikTok) take specific actions as well.


Well, they are being asked to make changes as well, which presumably prevents the damage.

It can be argued the changes don't go far enough so maybe "pay fine, reduce harm, but harm still permitted..."


The prosecutors representing the justice departments of states of California, Colorado, Kentucky, and New Jersey


For what its worth, if you believe Meta's product does do this damage, the other products which do similar things should also abide by it.

It helps simultaneously "level the playing field" but also protects consumers.

If the courts stated ruling is hinged on consumer protection then this seems fair. Any competitor which doesn't follow suit is likely opening themselves up to a similar lawsuit which would cost a similar order of magnitude of damages (so it is in the interest of the competitors to do take these actions)


It's basically contingent on the AGs suing those companies as well.

> and paying the same amount, split between the two companies.

So it's a bucket of crabs


The US is unique in its "cable" and telecom approach (maybe shared with some other Western countries).

In most APAC and African countries cell plans are extremely competitive and data is SUPER cheap. Same with streaming services.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: