The long-term value thesis of AI is much deeper than stock valuations, large model moats, chip suppliers, or business models around scalable parrots. It's nothing less than this: What happens to an economy and a society when you scale up qualitatively the cognitive power you can apply in every activity? (Side question: How can you make sure you (as a person, organization, or society) benefit from it?.)
Neither the unwinding of the yen carry trade nor the increased risk of a US recession nor tumbling stock valuations built around increasingly meaning-free uses of "AI" as the new "tokenized" have much to say about this. The impact of true AI — not the dream of AGI but the more interesting and pragmatic case of superhuman tools in narrow fields — dwarfs in scale and scope the ups and downs of a highly leveraged and flighty global financial system. If you believe the full transformative impact of a expanding computational frontier is still mostly unrealized (and the chasm between the math and the application is vast enough to make it clear) then what happens to NVDA or GOOG is, while not irrelevant to your short-term decision-making, uninformative about your long-term play.
(This isn't, by the way, an argument to invest when it's cheap. It's the observation that some shifts are so profound that the usual price signals fail. If you're working on the next qualitative generation of applied cognitive systems, the case for it is not an epsilon less true today than last week.)