>Myspace was bought out in 2005, and it had been dead for years before that.
Myspace's peak popularity (in terms of monthly unique pageviews) came in 2007. Yahoo was ahead of Google for several years, though I am struggling to find good data on the specifics there. Your understanding seems to be based in your own personal experience of those sites, which is fine, but I don't think it quite matches the big picture.
Your proposal that there won't be any new entrants is puzzling to me because, in the grand scheme of things, OpenAI and Anthropic ARE new entrants. The entire space is new. It feels like ChatGPT became public eons ago, but it hasn't even been 4 years. Imagine saying in 2008 that no photo-sharing site would ever surpass Flickr - it's been dominant for 4 years! Why would you go to something new when everyone's photos are already on there?
One absolutely enormous advantage new entrants have in the AI space is that they can start on third base. Anthropic had to spend billions and billions of dollars to get to Opus 4.6. A new entrant could distill its way into a model roughly on par with that far more cheaply, and spend its VC cash on things like marketing and price cutting rather than burning it on actually creating a good model. The existing players need to service mountains of debt and are locked into deals for compute based on outdated assumptions about efficiency. I'm not saying it would be EASY for a new entrant to waltz in, especially to the enterprise market. But it's not impossible especially on the consumer side: Meta was essentially a non-factor until Muse came out, and now most people think of Muse when they think of an AI agent (as opposed to Claude, which is just a chatbot in the public imagination, because that's what it was when it became popular).
i agree that they are, objectively, new entrants, but that does somewhat mischaracterize, in the same sense that calling Ford a new entrant into the car industry after they came out with the Model T. like... they are creating the market into which they are entering. ChatGPT galvanized this entire phenomenon.
i do agree that there is a good chance that one of these post- OpenAI/Anthropic competitors probably will get some sort of unanticipated foothold, if only because enough talent will drain out over time. what i push back on is the idea that being 90% as good as frontier models is actually good enough to matter in even the slightest. especially since we know that the "real" internal models are so much further advanced than what we get access to publicly.
i absolutely agree that it isn't impossible-- if the only thing stopping a new entrant is capital (as in: they have the trillion-dollar idea necessary to be competitive) that will absolutely never be a true impediment to a worthy competitor.
It's not even clear that the "frontier AI lab" is going to be the best, most profitable business model to evolve out of this tech. Especially if models are easy to distill, a new lab can start on 3rd base, and China open weights aren't far behind.
The winning business model may be 1 or 2 abstractions removed from "build a really good LLM model", and yet to be revealed in the marketplace.
Most people do not think of Muse as an the ai agent. Most think OpenAI or Claude. The people using Muse are using it as a facebook chatbot.
> Most people do not think of Muse as an the ai agent. Most think OpenAI or Claude. The people using Muse are using it as a facebook chatbot.
Yeah, I was with the case bring made, until that point. A very bizarre claim, especially given the questions around safety:
https://www.malwarebytes.com/blog/privacy/2026/10/metas-muse...