There are layers to this.
Yes, there is zero chance any of these investments are going to turn a profit.
But research and technology is fundamentally built around developing capabilities for long term power (soft or hard). That is WHY governments invest so much into university groups and research divisions at companies to develop features of interest. You are never going to make back the money that funded hundreds of PhD students to develop a slightly more durable polymer compound. But you will benefit because you now have hundreds of new graduates aligned with fields of interest AND a slightly better grip on your military grade sybian.
And, regardless of what people want to believe, AI genuinely does have some great uses (primarily pattern matching and human interfaces). And... those have very big implications both in terms of military capability and soft power where the entire world is dependent on one nation's companies for basic functionality.
Of course, the problem is that the "AI craze" isn't really being driven by state governments at this point. It is being driven by the tech companies themselves and the politicians who profit off of them. Hence why we are so focused on insanely expensive search engine replacements and "AI powered toaster ovens" rather than developing the core technologies and capabilities that will make this more power efficient and more feasible for edge computing.
And... when one of (if not ) the super powers is actively divesting itself of all soft power at an alarming rate.... yeah.
