Ask HN: What economic reasoning is there that bigCos have an advantage post-AI? | Hacker News Reader

Ask HN: What economic reasoning is there that bigCos have an advantage post-AI?

1 pointjppope5 comments

I've been following a lot of the discussion surrounding transformation in the market/world following the impact of LLMs (e.g. https://news.ycombinator.com/item?id=44892283) and there seems to be a forgone conclusion that large corporations will capture more value than small business or entrepreneurs and I'm struggling to follow the reasoning.

To me if individuals or small groups have the ability to use tools (LLMs) at greater leverage than larger groups, company size actually puts bigger companies at a disadvantage because of the overhead they are maintaining. I'm not excluding other market factors like brand or network effects, etc but if someone could explain to me the reason why capital has an advantage here if AI is commoditized not monopolized (the current trend).

Bonus points for any studies you can provide.

View on HN

People only mean that AI companies themselves (which are huge by necessity because of immense scale of compute required, and limited hyper-expensive, unique talent they use, although that is secondary). They are supposed to grab all the value from all the industries much like say, Google took out all profits and value from all traditional media, advertisers, retailers, virtually everything. So it will be like, 2-5 hyper large companies controlling majority of GDP. Which may be actually not that bad. Being in the driver seat, they may provide an alternative to 2 political parties - certainly a much saner alternative. Because they will be able to decide just about everything - and government won't be able to stop them because of their transnational nature, and very soon, will be unable to stop them by force as well - they may create a more sensible government by sidelining and/or co-opting traditional political system and make everyone happier. It might look a bit like Lebanon at times, though.

When people say that we are entering new Middle Ages, they might actually be cool.

Maybe but at this time it doesn't appear to be a "winner take all" market, its actually looking more and more like a commodity or utility market... There doesn't appear to be a moat or proprietary advantage to any providers (except at the hardware level) or anyone at the application layer. I've never once heard that someone would only use ChatGpt and could never use Claude. The improvements between model versions are minor at this point. How would someone win this market? I think its also a foregone conclusion that the only way to profitably monetize this market will be ads or to raise the price for inference in the future.

It's not like they will make money by selling their services. They will make money by being everything else. Like, "replace newspapers and news sites by AI content they generate, putting everyone in the news and media out of jobs".

It sounds scary as fuck, but let's see how it plays out.

I don't know exactly what you're referring to but if LLM is the thing that you're referring to as being the differentiator - then a company with a lot of capital can pay for a lot more LLM calls than one that is small.

Sure, but they still need to know what calls to make for new things. For old things someone can likely reproduce the things that they have built for less money, with less overhead now.

Reply on news.ycombinator.com