Frieren

the reason why I believe that all the american ai companies will inevitably lose the race is hubris plain and simple there is an uncanniness to all of them that makes them extremely untrustworthy they get a taste of the power of being the ai market leader and it corrupts them instantly and they lose all composure and strategic awareness we’ve seen it happen with all of them by now we saw it happen with openai after chatgpt where they thought they had already won, and then their organization imploded internally - Sam Altman went from likeable to becoming a smug villainous character that lost most of his public goodwill we saw it with Xai where elon initially promised to build open source models, and that he was building completely objective ai grounded in unbiased truth - and again as soon as Grok became just slightly decent he goes back on all his promises, no open source, grok becomes a sycophant machine programmed to praise everything elon does without any grounding in critical objective truth and we now see the inklings of it with Anthropic, initially they positioned themselves as being the ‘good guys’ amidst the elon and sama characters, claiming to do it all for the love of humanity and what not - but now as soon as they gained the lead you see the same patterns unfolding, a smug arrogance, an obsession with control, using fear campaigns to try and entrench themselves, etc it’s the same pattern every single time - fake humility and promises of benevolence and public good - as soon as they gain power it instantly corrupts their egos and they become controlling smug villainous characters that want to rule the world and that’s just from a philosophical pov from a technical and business pov it’s even worse for them american ai labs had the lead by far 3 years ago, 2 years ago, 1 year ago but today? - they’re barely holding on sure claude and gpt may hold a marginal lead for text and programming models, but chinese open source models are right behind them in terms of video models, China has the best by far, they won so hard in video that openai gave up and closed sora but we all know that text and programming is the most important so let’s look at it realistically yes gpt and claude are slightly better than the rest - but that’s really only of you’re a high end user willing to pay anything for a slight advantage in terms of token efficiency measured by performance divided by costs - they’re losing glm 5.1, kimi k2.6, qwen 3.6 - all of these models are chinese, open source, and cost a tenth of the price from the providers that claude and openai cost a coding max plan for using glm 5.1 costs less than a codex or claude max plan while providing maybe 1.5x as many tokens you basically get a model that’s roughly the same as opus 4.6 and gpt 5.4, but for less than half the price and then we also need to consider that anthropic and openai are currently lighting their money on fire - they’re subsidising costs and will continue to increase the prices and lower the rate limits for their models thus they can hardly afford to remain competitive meanwhile the chinese labs are at their heels, competing with them in model quality and out competing them in cost efficiency for users and we haven’t even seen DeepSeeks big release of the year yet everything considered China is currently winning and America is currently losing it’s obvious for anyone with eyes to see - you just have to know how to cut through the noise and analyze all the signals