Hacker News
points by skippyboxedhero 3 days ago | 0 comments
The long-term is considerably easier because in the short-term there are limits on supply so prices aren't really able to move in a way where balance will be found.

In the long-term, the marginal cost of a request is zero so prices will plunge and a lot of the fixed investment will get stuck. The difference between the cost of training a model/infra and serving a request from that model is a clear problem for economics.

What happens with Anthropic specifically is less clear because they, as every company in the same position in the last 150 years has done, are going to push for regulation. A lot of the risk has also been shifted onto neoclouds. The value of the customer relationship is, however, effectively zero. A core part of their strategy has been to build stuff out like CC to differentiate that but, ultimately, that won't work.

The other strategy they have employed is trying to get other companies to stop training new models. Again, seems unlikely to work.

There is no scenario in which they become Google search outside of government intervention. Obviously this is a key part of their strategy now but at best, this will only work in the US and Europe (I would actually say they are much further along in Europe where there is a strong tendency towards corporatism and centralisation of power, they could be "Europe's AI").

[–] jacquesm 3 days ago
The only thing regulatory capture in the USA will do is to improve the situation for the rest of the world to compete. I wonder how long it will take them to figure that one out, for all their smarts they seem to be incapable of reading a history book.
[–] 0xDEAFBEAD 3 days ago
This "regulatory capture" stuff mostly just exists in the imaginations of HN users anyway. This is just one reason why it never made sense as a strategy.

https://www.astralcodexten.com/p/ai-safety-is-not-a-corporat...

[–] petcat 3 days ago
> improve the situation for the rest of the world to compete

The risk is that USA and China will come together on a joint "AI non-proliferation" agreement in the same way that nuclear technology became restricted by the countries that already had the capability. This will lead to a de facto techno-colonization of the rest of the world just between themselves.

This was a topic of discussion at Trump and Xi's AI summit just last month.

[–] jacquesm 3 days ago
Neither Xi nor Trump are going to be able to make that stick. The cat is well out of the bag by now.
[–] petcat 3 days ago
Nobody cares about hobby models running on a 5 year old GPU cluster in the basement. This is about industrial-scale AI which, to date, only USA and China have accomplished.
[–] jacquesm 3 days ago
If you think that these are hobby models you are simply saying you haven't used them.

One of the largest legal services partnerships here in NL just launched their local DC so they can maintain confidentiality. Do you think they are 'running hobby models on 5 year old GPU clusters in the basement'? Or does a couple of racks of GB300's sound like maybe they know what they're doing after all?

[–] petcat 3 days ago
That is microscopic scale compared to what we are talking about. From an investor and "market" perspective, the only AI that matters is the one that's built into every iPhone and Google Android phone, every car, every TV, every smart appliance. The AI models that power ships, planes, and all kinds of robotics. The frontier AI that services the energy grid, and even the smart homes and very structures that we dwell in.

This is all on the horizon. We are only barely scratching the surface and right now it is largely hobbyist having fun with it, but we already see the potential of the AI revolution and where it is going.

[–] cman1444 3 days ago
It matters from the perspective of tech dependency.

If a medium/small country deemed it a threat to rely on US/China for AI capabilities, they can get 95% of the utility by banning all American and Chinese models and just using local ones.

This contrasts to something like oil or nukes, where if you're not one of the countries blessed with them, you have a serious power imbalance.

[–] pianopatrick 3 days ago
Seems to me that from an investor perspective, the AI models built into the cars and TVs and other things are likely to be open source. I think long term, due to the high cost of training these models, there is going to be a "Linux of AI". I.e. some model that lots of companies collaborate on and then use for free.

The economic logic is the same as that of operating systems. AI models are very expensive for one company to make, but lots of companies need them. So companies are better off collaborating on that side of things and competing on something else.

Economically, the AI models that require big servers with lots of GPUs in data centers should likely be more interesting to investors than the AI models that can run on local hardware.

[–] butlike 3 days ago
And then a random sun flare like in 1859 wipes it all out

https://en.wikipedia.org/wiki/Carrington_Event

[–] jacquesm 3 days ago
Ah, the True Scotsman of 'hobby labs'.
[–] qwytw 3 days ago
It just doesn't really make much sense. It's way harder to stop a company in a country which hasn't signed it (not that the congress ever would sign a random document Trump and Xi come up with regardless...) from developing a model than from building nukes. Also if American and Chinese labs stop development it would make it cheaper and easier for others to catch up.
[–] speleding 3 days ago
Although I agree inference will become cheap (probably not zero), it's quite possible that there is a moat they can build around understanding customer context.

With each query OpenAI and Anthropic get more context about their users. When I compare answers between AIs the ChatGPT answer often includes useful things it inferred from earlier conversations. Losing that context could be a meaningful barrier for a customer to switch to a cheaper service.

[–] jacquesm 3 days ago
> With each query OpenAI and Anthropic get more context about their users.

Not unless they break their own TOS. The temptation must be very large to do so and is probably why those settings keep mysteriously shifting back to 'yes, I give all my data to you to do with as you please'.

[–] sfblah 3 days ago
I don't think that's what GP meant. I think the point is maybe they could build a moat by "knowing you" in a way that makes you care about using Claude rather than, say, Gemini.
[–] jacquesm 3 days ago
I don't actually want them to know me. And if they build a file on me without telling me that's actionable under the GDPR.
[–] redanddead 3 days ago
Wild how even when humanity achieves a level of abundance there’s somehow a need from a few on wall st to create artificial scarcity.
[–] jacquesm 3 days ago
They can't help themselves. They all saw that movie and figure they can be stars too.