We can't rule out a new innovation that makes frontier models more relevant than deepseek in 6 months. Things evolve so fast.
The USA has the biggest, but there lies their disadvantage
In the USA building bigger, better frontier models has been bigger data centres, more chips, more energy.
China has had to think, hard. Be cunning and make what they have do more
This is a pattern repeated in many domains all through the last hundred years.
tant pis