[jsnell]

Yes, many people believe that, but it doesn't seem to be an evidence-based belief. I've written about this in some detail[0][1] before. But since just linking to one's own writing is a bit gauche and doesn't make for a good discussion, I'll summarize :)

1. There is no point in providing paid APIs at negative margins, since there's no platform power in having a larger paid API share (paid access can't be used for training data, no lock-in effects, no network effects, no customer loyalty, no pricing power on the supply side since Nvidia doesn't give preferential treatment to large customers). Even selling access at break-even makes no sense, since that is just compute you're not using for training, or not selling to other companies desperate for compute.

2. There are 3rd-party providers selling only the compute, not models, who have even less reason to sell at a loss. Their prices are comparable to 1st-party providers.

3. Deepseek published their inference cost structure for R1. According to that data their paid API traffic is very lucrative (their GPU rental costs for inference are under 20% of their standard pricing, i.e. >80% operating margins; and the rental costs would cover power, cooling, depreciation of the capital investment).

Insofar as frontier labs are unprofitable, I think it's primarily due to them giving out vast amounts of free access.

[0] https://www.snellman.net/blog/archive/2025-06-02-llms-are-ch...

[1] https://news.ycombinator.com/item?id=44165521

[lmeyerov]

I think you miss 2 big aspects:

1. High volume providers get efficiencies that low volume do not. It comes from both more workload giving more optimization opportunities, and staffing to do better engineering to begin with. The result is break even for lower volume firms is profitable for higher volume, and as high volume is magnitudes more scale, this quickly pays for many people. By being the high-volume API, this game can be played. If they choose not to bother, it is likely because strategic views on opportunity cost, not inability.

That's not even the interesting analysis, which is what the real stock value is, or whatever corp structure scheme they're doing nowadays:

2. Growth for growths sake. Uber was exactly this kind of growth-at-all-costs play, going more into debt with every customer and fundraise. My understanding is they were able to tame costs and find side businesses (delivery, ...), with the threat becoming more about category shift of self-driving. By having the channel, they could be the one to monetize as that got figured out better.

Whether tokens or something else becomes what is charged for at the profit layers (with breakeven tokens as cost of business), or subsidization ends and competitive pricing dominates, being the user interface to chat and the API interface to devs gives them channel. Historically, it is a lot of hubris to believe channel is worthless, and especially in an era of fast cloning.

[jsnell]

> High volume providers get efficiencies that low volume do not

But paid-per-token APIs at negative margins do not provide scaling efficiencies! It's just the provider giving away a scarce resource (compute) for nothing tangible in exchange. Whatever you're able to do with that extra scale, you would have been able to do even better if you hadn't served this traffic.

In contrast, the other things you can use the compute for have a real upside for some part of the genai improvement flywheel:

1. Compute spent on free users gives you training data, allowing the models to be improved faster.

2. Compute spent on training allows the models to be trained, distilled and fine-tuned faster. (Could be e.g. via longer training runs or by being able to run more experiments.)

3. Compute spent on paid inference with positive margins gives you more financial resources to invest.

Why would you intentionally spend your scarce compute on unprofitable inference loads rather than the other three options?

> 2. Growth for growths sake.

That's fair! It could in theory be a "sell $2 for $1" scenario from the frontier labs that are just trying to pump up their revenue numbers to fund-raise from dumb money who don't think to at least check on the unit economics. OpenAI's latest round certainly seemed to be coming from the dumbest money in the world, which would support that.

I have two rebuttals:

First, it doesn't explain Google, who a) aren't trying to raise money, b) aren't breaking out genai revenue in their financials, so pumping up those revenue numbers would not help at all. (We don't even know how much of that revenue is reported under Cloud vs. Services, though I'd note that the margins have been improving for both of those segments.)

Second, I feel that this hypothetical, even if plausible, is trumped by Deepseek publishing their inference cost structure. The margins they claim for the paid traffic are high by any standard, and they're usually one of the cheaper options at their quality level.

[lmeyerov]

I think you ignored both of my points -

1. You just negated a technical statement with... I don't even know what. Engineering opportunities at volume and high skill allow changing the margin in ways low volume and low capitalization provider cannot. Talk to any GPU ML or DC eng and they will rattle off ways here. You can claim these opportunities aren't enough, but you don't seem to be willing to do so.

2. Again, even if tokens are unprofitable at scale (which I doubt), market position means owning a big chunk of the distribution channel for more profitable things. Classic loss leader. Being both the biggest UI + API is super valuable. Eg, now that code as a vertical makes sense, they bought more UI here, and now they can go from token pricing closer to value pricing and fancier schemes - imagine taking on GitHub/Azure/Vercel/... . As each UI and API point takes off, they can devour the smaller players who were building on top to take over the verticals.

Seperately, I do agree, yes, the API case risks becoming (and staying) a dumb pipe if they fail to act on it. But as much as telcos hate their situation, it's nice to be one.

[jsnell]

I don't think I was ignoring your points. I thought I was replying very specifically to them, to be honest, and providing very specific arguments. Arguments that you, by the way, did not respond to in any way here, beyond calling them "[you] don't even know what". That seems quite rude, but I'll give you the benefit of the doubt.

Maybe if you could name one of those potential opportunities, it'd help ground the discussion in the way that you seem to want?

Like, let's say that additional volume means one can do more efficient batching within a given latency envelope. That's an obvious scale-based efficiency. But a fuller batch isn't actually valuable in itself: it's only valuable because it allows you to serve more queries.

But why? In the world you're positing where these queries are sold at negative margins and don't provide any other tangible benefit (i.e. cannot be used for training), the provider would be even better off not serving those queries. Or, more likely, they'd raise prices such that this traffic has positive margins, and they receive just enough for optimal batching.

> You can claim these opportunities aren't enough, but you don't seem to be willing to do so.

Why I would claim that? I'm not saying that scaling is useless. I think it's incredibly valuable. But scale from these specific workloads is only valuable because these workloads are already profitable. If it wasn't, the scarce compute would be better off being spent on one of the other compute sinks I listed.

(As an example, getting more volume to more efficiently utilize the demand troughs is pretty obviously why basically all the major providers have some sort of batch/off-peak pricing plans at very substantial discounts. But it's not something you'd see if their normal pricing had negative margins.)

> Engineering opportunities at volume and high skill allow changing the margin in ways low volume and low capitalization provider cannot.

My point is that not all volume is the same. Additional volume from users whose data cannot be used to improve the system and who are unprofitable doesn't actually provide any economies of scale.

> 2. Again, even if tokens are unprofitable at scale (which I doubt),

If you doubt they're unprofitable at scale, it seems you're saying that they're profitable at scale? In that case I'd think we're actually in violent agreement. Scaling in that situation will provide a lot of leverage.

[lmeyerov]

> But why? In the world you're positing where these queries are sold at negative margins and don't provide any other tangible benefit (i.e. cannot be used for training), the provider would be even better off not serving those queries. Or, more likely, they'd raise prices such that this traffic has positive margins, and they receive just enough for optimal batching. ... But scale from these specific workloads is only valuable because these workloads are already profitable

I'm disputing this two-fold:

- Software tricks like batching and hardware like ASICs mean what is negative/neutral for a small or unoptimized provider is eventually positive for a large, optimized provider. You keep claiming they cannot do this with positive margin some reason, or only if already profitable, but those are unsubstantiated claims. Conversely, I'm giving classic engineering principles why they can keep driving down their COGS to flip to profitability as long as they have capital and scale. This isn't selling $1 for $0.90 because there is a long way to go before their COGS are primarily constrained by the price of electricity and sand. Instead of refuting this... You just keep positing that it's inherently negative margin.

In a world where inference consumption just keeps going up, they can keep pushing the technology advantage and creating even a slight positive margin goes far. This is the classic engineering variant of buttoning margins before an IPO: if they haven't yet, it's probably because they are intentionally prioritizing market share growth for engineering focus vs cost cutting.

- You are hyper fixated on tokens, and not that owning a large % of distribution lets them sell other things . Eg, instead of responding to my point 2 here, you are again talking about token margin. Apple doesn't have to make money on transistors when they have a 30% tax on most app spend in the US.

[lmeyerov]

Maybe this is the disconnect for the token side: you seem to think they can't keep improving the margin to reach profitability. They are static and it will just get worse, not better.

I think deepseek instead just showed they haven't really bothered yet. They rather focus on growing, and capital is cheap enough for these firms that optimizing margins is relatively distracting. Obviously they do optimize, but probably not at the expense of velocity and growth.

And if they do seriously want to tackle margins, they should pull a groq/Google and go aggressively deep. Ex: fab something. Which... They do indeed fund raise on.

[solarkraft]

> Engineering opportunities at volume and high skill allow changing the margin in ways low volume and low capitalization provider cannot

Everything depends on this actually being possible and I haven’t seen a lot of information on that so far.

DeepSeek‘s publication suggests that it is possible - specifically there was recently a discussion on batching. Google might have some secret sauce with their TPUs (is that why Gemini is so fast?)

And there are still Cerebras and Groq (why haven’t they taken over the world yet?), but their improvements don’t appear to be scale dependent.

Speculating that inference will get cheaper in the future might justify selling at a loss now to at least gain mind share, I guess.

[lmeyerov]

There is an interesting subtlety to what you are saying -

* Inferencing is getting cheaper: 40-80% / yr in hardware + 40-80% / yr in software. Annual numbers will change, but for next 5 years, seems fine to assume continued improvement at meaningful levels.

* Subtlety 1: What % of that improvement is only accessible to large providers with bigger workloads and bigger capitalization? Think more opportunities for batching, specialization, bulk purchasing, ... . That edge is where their ability to both underprice others + have a profitability margin here vs being a loss leader. A lot of the annual hw/sw improvement is commoditized for low-scale. Eg, hyperscalers pay 10% less to Nvidia than the next tier of smaller buyers, and I'm unsure of how to think of the margin benefit of custom hw

* Subtlety 2: if things keep going well, there may be a limited window around wirkloads "tricky enough that only they can do it" , eg, if commodity + oss using 2nd tier providers are "good enough" and also reach low COGS, so it's just regular data center stuff, all this becomes unclear again. I can likewise imagine that becoming true of the wider use case of Text, while image/video/audio ends up more specialized

[what]

There are more factors to cost than just the raw compute to provide inference. They can’t just fire everyone and continue to operate while paying just the compute cost. They also can’t stop training new models. The actual cost is much more than the compute for inference.

[jsnell]

Yes, there are some additional operating costs, but they're really marginal compared to the cost of the compute. Your suggestion was personnel: Anthropic is reportedly on a run-rate of $3B with O(1k) employees, most of whom aren't directly doing ops. Likewise they also have to pay for non-compute infra, but it is a rounding error.

Training is a fixed cost, not a variable cost. My initial comment was on the unit economics, so fixed costs don't matter. But including the full training costs doesn't actually change the math that much as far as I can tell for any of the popular models. E.g. the alleged leaked OpenAI financials for 2024 projected $4B spent on inference, $3B on training. And the inference workloads are currently growing insanely fast, meaning the training gets amortized over a larger volume of inference (e.g. Google showed a graph of their inference volume at Google I/O -- 50x growth in a year, now at 480T tokens / month[0])

[0] https://blog.google/technology/ai/io-2025-keynote/

[ummonk]

That's all the more reason to run at a positive margin though - why shovel money into taking a loss on inference when you need to spend money on R&D?

[brookst]

I heart you.

Classic fixed / variable cost fallacy: if you look at the steel and plastic in a $200k Ferrari, it’s worth about $10k. They have 95% gross margins! Outrageous!

(Nevermind the engine R&D cost, the pre-production molds that fail, the testing and marketing and product placement and…)