i have recently noticed among the ai companies, there's a clear distinction between winners and losers.
the winners, as of february of 2026, are anthropic, followed steadily by google. and then there's fucking openai.
anthropic sounds like a research company but they aren't trying to build a normal consumer product. there must have been a pivot sometime for the past year or so that they realized if they can achieve monopoly in AI coding, they've won, and that non-demanding users (in terms of their API) simply aren't as profit making as the developers. this is the only reason why ai benchmark matters because developers look at SWE bench, try the top one to experience how good or bad it is, and if it's the former, they switch. and for the past few months the winner in the benchmark game has always been anthropic.
onto google. google doesn't care about their cash burn because of the following: they own android (which is free on the surface, but they own the google play store fee, which is like 70-80% of worldwide smartphones), chrome + google (the ads business), youtube (again, ads...), and so gemini under deepmind is just one of their small and not extremely consequential bets that turned out to be an extremely good bet. they bet on their raw capital power. anthropic perhaps competes in modeling (aka training) in a smarter way, but if google improves their engineering they could easily claim back the first spot.
onto openai. i said this before but they lose money everday, and now desperately try to put ads on their only product, chatgpt. other companies don't do that because they actually make money and they make it elsewhere. openai is hopeless. their ceo is a scam (loopt was a scam), and their company will definitely fail. the original cofounding team has collapsed entirely, leaving the only nontechnical scammer in the center of it all, and this colossal will eventually fall, because they couldn't compete in model engineering (against anthropic), or having as much cash as possible (against google/deepmind).
the lesson is to always make money first, then by extension, optionally consider R&D.