Depending on how you feel about Jevon's paradox, you've had either more or less work to do since the US Government's embargo of Anthropic's Fable model.
More, because you have to do the same stuff you were trying to get done using lesser models, or less because the explosion of new stuff you were going to do with a more powerful model did not eventuate.
While kind of in the first camp, I've nonetheless found some time to think about what this precedent might mean for the Australian AI landscape and who exactly has how much leverage to deny what to whom.
For example, the US can deny Mythos/Fable-class models to the rest of the world, including China, as evidenced by recent events. But then, if this were to escalate, China could deny hardware, and therefore frontier AI, to basically everyone by blockading Taiwan (no invasion needed - the Iran conflict showed that it is very difficult if not impossible to keep commercial shipping lanes open against a serious blockade even by a smaller, low-tech navy).
What about Europe? You've probably heard of ASML - the Dutch company that produces basically 100% of state-of-the-art lithography machines used in silicon chip production.
Sure that there are more levers I wasn't aware of, I had Claude Code run some research on this and put together an explainer. The AI production supply chain is a sequential series of steps that involves Africa, China, the Netherlands, Germany, Japan, South Korea, Taiwan and the USA in a way that basically any one of them withholding their part would break the whole thing, and it would take decades for anyone else to domestically replicate it.
It probably won't surprise anyone that Australia is, for the moment, nowhere in the picture.