Analysis
Hermann Hauser has earned the right to have an opinion on whether AI is a bubble. He co-founded Arm, the chip architecture that ended up underneath essentially every mobile device on Earth, and has spent the decades since backing European deep-tech startups through Amadeus Capital. So when he told CNBC this week that AI 'is a revolution that will create more value than probably any other technology revolution that we've ever seen,' I think that's worth taking seriously -- and I think his more specific follow-on point is the one most of the bubble debate is missing entirely.
Hauser's actual framing isn't 'bubble' or 'no bubble.' It's that the revolution is real, but 'the prices are not all real' -- and, critically, that the biggest names aren't the fragile part of the market. That's a genuinely different claim than the binary most AI-valuation commentary defaults to. It says: don't confuse skepticism about a specific valuation multiple with skepticism about the underlying technology, and don't assume the companies getting the most headline attention are automatically the ones most exposed if sentiment turns.
“That's a genuinely different claim than the binary most AI-valuation commentary defaults to.”
I'd extend that logic directly to what Pulse has been tracking all month. Anthropic reported preliminary Q2 revenue above $11.5 billion this week, with growth that looks nothing like a bubble by any reasonable revenue-multiple standard. Meanwhile, a wave of infrastructure rounds -- nuclear reactors, home batteries, automated factories -- are being priced on capacity that doesn't exist yet, which is a genuinely different and riskier kind of bet than a revenue-backed AI-lab valuation. Hauser's distinction between 'the revolution is real' and 'the prices are not all real' maps almost exactly onto that split: the biggest, most revenue-proven names are the least likely candidates for a violent repricing, while newer infrastructure bets priced purely on future capacity are where the real fragility sits.
Room for disagreement: Hauser also runs a venture fund with obvious incentive to keep AI capital flowing rather than sound a genuine alarm, and 'the revolution is real but some prices aren't' is a comfortably hedged position that lets him be right regardless of what actually happens next -- if AI valuations crash, he warned about mispricing; if they don't, he affirmed the revolution. It's also worth noting his second warning, about Europe risking becoming 'a technology colony of the U.S.,' is a distinct argument about sovereign capacity and dependency, not really about valuation at all, and conflating the two in a single soundbite makes his overall message easier to agree with in the abstract than it would be if each claim were evaluated separately.
For founders and LPs, the actionable version of Hauser's take is this: stop asking 'is AI a bubble,' which is a question with no useful answer, and start asking which specific layer of your portfolio is priced on proven revenue versus unbuilt capacity. That's a diligence question you can actually act on.
