When most people think of an AI bubble, they imagine frothy tech stocks, inflated valuations and investors throwing their money at anything with the words “AI” in it. But the guy with the most prominent job in investment research and strategy at Goldman Sachs is looking at that same landscape — and coming to a completely different conclusion. The bubble, as he sees it, isn’t where headlines would point. And it is that disparity between perception and reality that explains why Wall Street has begun taking his warning seriously.
The strategist’s case begins with an obvious observation: The biggest beneficiaries of AI on the stock market — chipmakers, cloud powerhouses and infrastructure players — are not acting like a bubble at all. Their revenues are booming, they’re in demand, and their customers aren’t buying hype. The kind of compute power, hardware and tools they actually need. By conventional standards, these stocks are hardly speculative manias. They’re growth on the back of real revenue.
So if it’s not sitting in tech stocks, where is this bubble? For him, it lies in expectations. Not of the share price variety, but of promises and marketing and that breathless Article of Faith that AI will change everything overnight. He believes non-tech companies — consumer, retail, healthcare, financial services and manufacturing companies — are pushing pell-mell to prove they're "AI-ready" even when their adoption is still fledgling or experimental or years from scale.
Read through earnings calls and you’ll sense it immediately. Executives speak about AI as if it’s a magic button. That half-baked pilot projects are trotted out as if they’re a long-term strategy. The savings are projected before the tools are in place. Whole industries are throwing A.I. around as a hype bomb, not because they’ve produced anything of substance yet. And Goldman’s Gilt (yes, that can be the nickname) says this is more of a 1999-style bubble than Nvidia’s stock chart would ever make you believe.
He doesn’t say AI is not transformative. He’s saying the timeline is being twisted. Real deployment takes time. It’s not as though you can just flip a switch and integrate AI into logistics, medical systems, compliance, insurance and manufacturing. But markets are behaving as if the world will flip a switch and unlock deep productivity overnight. That pressure is beginning to force companies to overpromise — just the thing that inflates bubbles in the first place.
He highlights something subtle but telling: most of the companies that are bragging about AI aren’t yet delivering on actual margin improvement. Their labor costs aren’t dropping. Their workflows aren’t transformed. Your efficiency numbers aren't that different. What they are making in the meantime is narrative — and narrative is what inflates bubbles long before the fundamentals show up.
His biggest worry is that investors are interpreting every corporate utterance of “AI integration” as evidence of future profits, even if the underlying business has no relation to automation or machine intelligence. From the perspective of an observer, it has something in common with the dot-com era, when companies rebranded themselves around a buzzword that had yet to prove it could salvage or revolutionize their business models. And when what you’re expecting ramps up faster than the delivery, disappointment is always close behind.
And that is why he thinks the real bubble isn’t financial — it’s psychological. A herd belief developing at a faster rate than the technology is maturing. The danger is not that tech giants will disappear. It’s not that scores of non-tech companies have sunken costs into an AI-driven future they may never truly achieve. And futures like those suck, because then shareholders will begin to wonder if AI adoption was always more marketing than transformation.
Still, his outlook isn’t bearish. He is not predicting a crash, only a recalibration. He believes the companies actually constructing AI infrastructure will continue to rule, because the demand is real. I wonder whether tech companies attempting to stretch the AI narrative to flatter investors will face tough questions at some point if the productivity miracles don’t show up.
The message he’s trying to convey to investors is straightforward: bubbles don’t build around technology that already works. They cluster around the promises that have yet to be challenged. And the loudest promises aren’t being made by Silicon Valley, at least not yet — they’re coming from industries that hope to prove that they won’t be left in the dust.
Of course, the irony is that eventually AI will remake the economy. Just not as quickly as corporate storytelling would have it. And when the hype fades and realism takes hold, the companies that didn’t overhype their progress could end up looking shrewder than those who are screaming most loudly today.
