EXCERPT In early 2025, observers in China may have wondered why many Western AI-watchers were so surprised to see a Chinese AI company enter the rarified list of leading open weights foundation models. But this surprise was understandable: the ascendancy of DeepSeek called into question so many presumptive Western narratives about what AI in China is and may become. This was not Skynet AI, but Street AI: fast, cheap, good, and easy to use, and it made the supposedly ‘open’ Western way of doing things look downright bloated and locked down in comparison. Perhaps more importantly, the arrival of DeepSeek suggested that the steering of large models toward open multilingual futures might—via the hypercoupling of models, training techniques, inference subcul- tures, prompt cosmologies, and application wrappers and scaffolds—delay or reverse the apparently imminent decoupling of the West and China driven by their antagonistic geopolitical positioning. Just as life finds a way to evolve, so does technology—because technology is a kind of life, and it will follow an evo- lutionary path that is irreducible to its instrumentality for human ends. If artificial superintelligence kicks in sooner rather than later, replicating itself and terrafor- ming its host planet toward full Kardashev I, then who patrols the South China Sea will probably be the least of our concerns. With the memorable DeepSeek panic in mind, my ‘apology’ to the Chinese reader is an attempt to explain why the response in the West was one of surprise and discombobulation. Below I will summarise what I see as some persistent Western habits of thought—habits that may seem odd because, seen from some distance, they are. …