- AI companies are stupidly overvalued, but that doesn’t mean AI itself is overrated
- Underlying AI technology is real and transformative, and will rise from the ashes of market hype and man-child CEO tantrums
- The true AI revolution is taking place in an industrial world that investors and analysts don’t understand
Here’s the advice I gave to optical switch startups 25 years ago:
“Picking the right VC is crucial. Try to find one that’s already funding a few service provider startups. That way, they can get you together and broker a fictional ‘contract’ showing that you’ve sold 50-million dollars-worth of your non-existent optical networking product to their non-operative optical networking service.” —Saunders, Light Reading March 2000
This is the same virtuous circle/circle-jerk model described in Bloomberg’s recently published AI infographic [below], showing members of the U.S. AI vendor ecosystem circulating revenue among themselves.
Bloomberg’s model has split opinion. In one camp, the Billy Boosters, who think it depicts a healthy ecosystem operating in “flywheel” mode, throwing off value/magic beans for all.
In the other, those who are convinced that it’s a portent of doom - just like the optical networking shenanigans that I laid out 25 years ago. These neggy nellies see an indicator of a classic bubble market designed to cynically hyper-inflate the valuation of hyperscaler tech stocks, and that will, at some point, burst – crashing the entire U.S. economy.
You’re with stupid
I’ve been in “camp AI bubble” since last year, but with a significant caveat.
Quantitative evidence for the bubble theory abounds. Nvidia’s brief dalliance with a $5 trillion valuation at the end of October contains an obvious logic bomb. The chipmaker’s stock is currently reaping the benefits of a right place, right time-incepted near-monopoly. But its first mover advantage, like its chips, will not last forever. They need replacing every three years at most. And by the time they expire, established companies such as AMD, Intel, Qualcomm, Google, Amazon, Microsoft, Meta Platforms, and Huawei will all be offering proven alternatives, alongside numerous new entrants like Groq, Cerebras Systems, Graphcore, Sambanova, and Tenstorrent.
Similarly, OpenAI’s commitment to spend $1.4 trillion on infrastructure against revenues of $20 billion, or Palantir’s trailing 12 months Price-to-Earnings ratio of 690 (hahaha) are equally obvious red flags – made yet redder and flaggier by their executives’ puerileresponses to anyone who dares question their greed narratives.
This is the textbook behavior of Peter Pan, man-child CEO’s (and at least one President) who can’t handle the truth.
Reasons to be cheerful
So why do I not feel thoroughly dispirited about the pending AI bubble collapse?
Because not all bubbles are created equal. The AI bubble is much closer to the optical networking bubble/burp of the turn of the Century than the subprime mortgage apocalypse of 2008. The fiber-optic market eventually rose like a Phoenix from the ashes of the optical networking collapse. It turned out the demand really was there; it was just the timing and valuations that the industry biffed.
Same deal with AI. Of course we’re in a bubble. But that doesn’t mean AI isn’t a transformative, world-changing technology; it will just take longer to manifest its success, après le deluge, and it will do it in ways that have little to do with current market foci.
Big Tech execs are successfully over-rotating the US media, government (sic), analyst fanboys, and investor sheeple towards the AI data centres and GPU chips at the core of the cloud. But those are just factories – the AI means of production, if you will.
The true potential of AI – and 5G, and automation, and robotics - lies beyond the edge of traditional telecom networks, in the heavy industry hinterlands that make up 75% of the world economy. It’s here that the next great industrial revolution is emerging. It’s the undiscovered country where fortunes will be made, and commercial and national economic legacies will be carved for the rest of this Century - long after the effects of the current AI bubble have faded.
Sweat the details
How should carriers, vendors, and investors navigate the confusing AI landscape?
First, do your own homework. Use your eyes. Talk to your network. I’m in the outback in Western Australia right now, visiting a mining company, Newmont, which is using AI to autonomously operate a fleet of 41 150-tonne mega-trucks over a private 4G/5G network from Ericsson. That tells me Industry 4.0 is real, and I’m right to be excited about it. If you can’t work out why you should be excited about AI in your business, maybe you shouldn’t be – yet.
Second, stop listening to AI vendors. Sam Altmann isn’t the magus of artificial intelligence. He’s just the guy who made it accessible to regular Joes--he’s literally the AI chatbot guy. Alex Karp isn’t an AI savant; he’s just a creepy asshat who gets off on boasting about using AI to kill people. And both are motivated by greed and ego.
Third, take any media or analyst guidance on AI with a bag of salt. Or do the opposite of what they’re recommending (sometimes just as good). None of the telecom analysts pointed out the flaw with the carriers’ original 5G business case (more bandwidth doesn't equal more revenue); 958 of 960 Silicon Valley VCs passed on Google’s business plan; and financial pundits never call the market crashes. These are not the people you should be taking advice from on AI strategy or investment. These are some other people.
Steve Saunders is a British-born communications analyst, investor and digital media entrepreneur with a career spanning decades.
Op-eds from industry experts, analysts or our editorial staff are opinion pieces that do not represent the opinions of Fierce Network.