I’ll say it. I don’t care: The AI hype machine’s starting to look a lot like the blockchain delirium ca. 2017. And like countless historical examples of the same phenomenon, merely saying you’re going to invest more in a technology at some point down the road is enough to catalyze huge market cap gains in the here and now.
I want to be clear: I’m a believer in the promise of AI across all sorts of applications. I own Nvidia, and not just in index funds. I’m the furthest thing from a naysayer and I’ll be the first to emphasize that what makes “this time different” — dare I employ that oh-so-dangerous phrase — is that the companies involved in the frenzy aren’t just good companies, nor are they merely great companies, they’re the best companies the world’s ever known, bar none.
One of those companies is Alibaba, and as discussed here on at least two occasions over the past month, they’ve got it goin’ — I don’t know what “it” is exactly, but whatever you want to call it, Alibaba’s workin’ it. Have a look:
The rally’s damn near a 90-degree angle — like gold’s record run.
With Wednesday’s ~10% advance, the monthly gain for the Hong Kong-listed shares is — and try not to laugh — 50%. That’s the most for any month in the near six-year history of the secondary listing.
Things started cookin’ in earnest when the company’s latest earnings release revealed accelerated AI-assisted cloud revenue. Around the same time, reports indicated Alibaba’s developing its own AI chips. The combination of those two news items drove the shares to their largest single-session gain in years and the stock hasn’t looked back since.
Have a look at that succession of weekly gains: 14%, 14.5%, 5.3% and 9% already this week.
So, what happened Wednesday? Why’d the stock pop again? Because Eddie Wu opened his mouth at a developer conference. “Demand for AI infrastructure far exceeded our expectations,” he said, adding that in order to keep up, the company intends to boost its own capex plan beyond the CNY380 billion tipped earlier this year.
That’s it. All it takes these days to drive enormous gains for enormous companies is for a somebody (and Eddie’s a somebody) to tip more AI capex spending.
I’m sure the “techies” among you can explain why the totality of Wu’s remarks justified Wednesday’s rally. But I’ve been doing this for a pretty long time now — which is weird to say, not to mention kinda depressing — and I can tell you having seen this stage of the hype cycle before that it’s a perilous juncture.
To reiterate: I think AI’s probably just as transformative as people like Jensen Huang insist, and as such I don’t doubt the bull case. Even if I did, I’m certainly not dumb enough to stand in front of this locomotive.
But merely mentioning your intention to spend an unspecified amount of additional money on a technology at an unspecified future date isn’t worth 10% on a company the size of Alibaba. It’s just not. That doesn’t make any sense.
All that was missing on Wednesday was Alibaba touting some manner of new business relationship with Nvidia. Oh, wait………
“Where’s the beef?” I am way more skeptical about the profits listed companies will generate from AI. Both providers and users. Make that large users, not the 74% who use it for personal things at this time.
But I can understand some of the enthusiasm from this announcement because it references actual customer demand for their products. Now we don’t know if it is demand from large paying customers, but it would be very reassuring if it was.
From the center of the labor substitution maelstrom I can report that AI makes experienced software developers 2-3x more productive, at minimum, and we are striving for 10x with the usual degree of techie fervor and neophilia.
Tech is, as you know, famously navel-gazey and tends to see itself as the world’s premier economic output. Let’s control for that. Let us also accept that the AI success stories vis a vis software are probably hype: AI doesnt replace programmers entirely; it costs productivity when used injudiciously, and vibe coded applications depreciate at literally 100x the rate of human-involved creations. So we’re clearly in an AI hype bubble regarding tech.
Nevertheless, an N00% productivity gain cannot be trifled with and for a cautious welder of these tools, we would see a net gain in economic productivity even at 5x the price. Thriftiness is easy to achieve with proper context engineering and divide-and-conquer agent flows.
So: when the bubble bursts, where will inference be priced relative to today’s deeply subsidized prices? And how will that impact the various commercial and personal-use market segments that are currently amenable to AI?
Let us take 5x price expansion as a worst case. The casual personal-use market would dry up or switch to cheap substitutes. Excepting search (which can be highly optimized, and where many would likely pay a premium or submit to adver-tokens in the output stream) I don’t see a huge market for $100 /mo personal productivity tools.
Yet in software, CRM, law, and a few other areas, the core market will remain.
Let’s naively assume that 80% of TAM evaporates under this 5x price increase. What will that do to the share price of our precious tech titans? My economic modeling abilities aren’t up to this task. Maybe I’ll ask Claude to help!
What about education? I have read that the right kind of teachers, with AI tools, can significantly improve the quality of education. Exciting stuff.
If it is successful in lower education, what are the odds that it will be used to reduce the number of teachers so as to allow property tax cuts? Over 90? That stuff has to pay its own way.
There is often a difference between a great company and a great stock.
Although it appears that the “suite” is limited, remember …
“China bans its biggest tech companies from acquiring Nvidia chips, says report — Beijing claims its homegrown AI processors now match H20 and RTX Pro 6000D
News
By Jowi Morales published September 17, 2025
This could be bad news for Nvidia.”
Anything can happen once they have the tools onboard.