Sell the IPO?

Bubble Complications

“The first is the extreme brevity of the financial memory. In consequence, financial disaster is quickly forgotten.”
– John Kenneth Galbraith, A Short History of Financial Euphoria

When things work for long periods of time, with no major periods of doubt in between, it's easy to get lulled in a false sense of security.

John Kenneth Galbraith referred to the extreme brevity of the financial memory, and I was reminded of that quote as I was walking the sunlit streets of London.

If you haven't read any of Galbraith's books, I would recommend you at least read A Short History of Financial Euphoria and The Affluent Society.

Scrolling FinTwit these days, you get a sense of perma bullishness that borders (or tramples) on arrogance.

Confirmation-Bias Investing 🙄

It's easy to construct arguments that promote a certain thesis – when that thesis is working. And when things start to turn a little bit, you start to change your tone, hedging the downside... but hedging the upside too. Whatever happens, you're a winner.

The trend is with you, the market is with you, the world is with you. And to be honest, so is FinTwit!

Putting abstractions to the side, what happened in Crypto? Anyone who was critical of Crypto and Crypto Treasury Companies was attacked and labelled as someone who doesn't get it, at the very least.

Have fun staying poor, they used to say...

This Time is Different

Of course you can't equate Crypto with AI – the latter actually has applications, adoption and is already solving real-world problems. RIGHT?

Yes, but scratch the surface...

The success (and hype) of AI is shaping our expectations of what comes next. Those expectations drive investment decisions, and corporate behaviour. The resulting actions then alter the very reality investors are trying to anticipate.

Perception and reality reinforce each other. The process is said to be reflexive.

This is what Citrini missed in his tweet last month, trying to argue that AI is not a bubble, when he said: "In the bubble, the reality is driven by the market - not the other way around."

Citrini here misses that in a bubble – the market is neither driven by reality nor is reality driven by the market. In a bubble, reality "drives" the market (i.e. perception) and the market (imperfect perception) drives reality, in a reflexive feedback loop.

That's the reflexivity that everyone talks about but doesn't really understand.

Reflexivity in AI

In this AI cycle, we could say the underlying trend is the proliferation of AI technology and its effects on human productivity. It's undeniable that this is happening, IRL.

This underlying trend (i.e. reality) can be quantified by usage of AI: either by tokens consumed, by real-world use cases adopting it, problems solved with AI and other metrics deemed to be real-world metrics.

The perception of the underlying trend (i.e. perception of reality) is measured in stock prices, general announcements, mega deals like Stargate or OpenAI x AMD and other strategic partnerships, AI features announced and perceived progress on things like AGI etc.

-> It's measured in all the things we fallible humans use to measure it.

Get to the Gist

There's no doubt that there are reflexive connections at play in AI right now.

In our Biden's Techno-Imperial Cycle piece from 2024, we created a flowchart that depicted the boom/bust process of the Techno-AI cycle.

In short, a massive capex cycle was unleashed when ChatGPT was launched and the world suddenly realised we needed more compute.

We embarked on a compute race that is still ongoing but that has notably evolved...

...Since then, the compute race has fully transformed into an AI arms race – with Big Tech players fighting an existential war between themselves – and rightfully so.

Why? Only the Paranoid Survive!

For Big Tech, missing out on AI could mean failure over the long term – no matter how big or dominant you are.

In the short-term, being *seen* to be losing on AI means your valuation gets cut at least in half – and management doesn't want that.

For Apple: being unable to integrate AI sufficiently into their products means someone else will do it, and slowly take market share.

For META: being unable to offer sufficient AI to their billions of users means they will gravitate to other platforms that have the offering. TikTok is already competing with Instagram, and who knows what Apple x Google will cook up anytime?

For Microsoft: Microsoft has already lost Social, Search and Smartphone. To an extent they are still around because of their massive moat in software – which prints cash. Not scaling in Cloud and AI means they remain a software company watching its moat slowly erode every year.

What's stopping Apple from launching more affordable computers and going mass market? Oh that's right, they already started doing that.

For Google: Search leader and dominant in their consumer tech suite of products but under threat from losing traffic on the interface layer.

The threat is simple: google.com was our internet headquarters. Now AI platforms are taking some of that share. This is a long-term threat for Google that they need to defend against.

Take the recent Siri AI flop by Apple – it got a lot of negative headlines, and so did Google, META and other Big Tech in AI-related launches.

But in the time of AI...

"It's better to launch and flop, than to not launch at all."

This is a major force driving the boom: major players are squeezed to do something. In many ways, they are forced buyers.

Remember the attacks against Apple management because they weren't doing enough in AI? Narratives are fickle.

But that “do something” anxiety is starting to reverse as investors grow less willing to reward AI at any cost.

We are slowly transitioning from AI at any cost to show me the economics.

A few things that are becoming obvious...

..and that I think will feed the narrative for a sustained decline in Tech:

On the picks and shovels side, deals are being signed but not being delivered
AI aims to solve problems using brute force and ever more compute
But it's becoming obvious that Elegance is the way, not more compute
The upcoming OpenAI and Anthropic IPOs have skewed reality and fitted it into a convenient narrative for these companies.
Conflicts of interest!
Old Mantra: Consume compute at all costs
New Mantra: Consume compute to create value
"Cost is the ultimate differentiator."
And more.

May the IPO be with you

The simple point is: once these companies are public, their motivations start to change.

It's no longer window dressing to raise the capital and keep the hype going...

It shifts to managing the cash they raised, continuing to invest in the business and trying to succeed for the longer term – in a fiercely competitive environment.

This means many of their pro-cyclical moves in the marketplace start to reverse, affecting everyone else in the stack.

Things then start to unwind 🔄

It can be said that the pre-IPO becomes a strange attractor that pulls all things up – but post-IPO the attractor loses its force, and gravity starts to take hold.

If I could offer a parallel to this strange attractor setup – think MicroStrategy and Bitcoin when the former was buying up the coin.

The back side of the cycle was when MicroStrategy stopped buying the coin 😬

Which brings us to the punchline...

Sell the IPO?

The ChatGPT moment initiated this whole AI boom process.

Today, growth has already slowed, and feature release after feature release has exhausted the pool of sensational new breakthroughs and feature releases!

Its archenemy, Anthropic, used every trick in the book to hack growth and surpass OpenAI in traffic, this tweet from Michael Fritzell shows traffic evolution for Gemini, ChatGPT and Claude.

Website traffic: ChatGPT has peaked pic.twitter.com/xlwvFHmIW5
— Michael Fritzell (Asian Century Stocks) (@MikeFritzell) June 8, 2026

We see how Gemini and Claude caught up and even surpassed ChatGPT – and if you think this can't happen again, even with new AI, you will be surprised.

AI lab traffic is like a wrecking ball, switching costs are very low and usage will continue to move to and fro – until the shifts taper down and settle.

None of these players will dominate the market – competition will remain fierce and perceived moats will continue to be fiercely attacked.

That doesn't bode well for OpenAI and Anthropic at trillion-dollar valuations.