Germanium Musing
An Al Jazeera article asks whether the US can win the AI Cold War with China.
But the question assumes that winning means the US polishing the current turd better than China does.
An interesting turd but a turd nevertheless.
The entire argument revolves around GPUs, data centres, chip restrictions, bigger models, who can throw the most compute at essentially the same technology, or who can give away more free AI services before running out of cash.
That assumes today’s AI architecture is basically the technology of the future.
It won’t be.
We are at the germanium transistor stage of AI. OpenAI is Philco. Anthropic is Sylvania.
Household names, not.
The US need not worry. They have a proven innovation engine in Silicon Valley. The US innovation system is the world’s innovation system. It attracts all the talent and has the most risk capital. Plus it doesn’t rely on government intelligence (sic).
And the Chinese, well they never had an enlightenment. Sneaky following won’t cut it.
If someone insists on treating AI development as a US-China contest, they are probably watching the wrong race.
