The Bubble Nobody Is Questioning

We remain worried about two things that usually travel together near a top: rising debt, and the financial chicanery on open display in private credit. Above all of it sits the AI bubble. Bubbles are normal at this stage of a cycle. The sheer size of this one is not, and that is the worry.

Look at the claims being made, and waved through. The CEO of Anthropic has again said AI will cure cancer within about five years, the second time that promise has been floated, and medical researchers call the timeline fantasy. A prominent backer says he would not be surprised if Anthropic is the only AI company left standing in a decade. Mark Zuckerberg, who now faces accusations that his products are built to hook children, has published an AI manifesto promising abundance for everyone. Elon Musk says money itself will soon stop mattering. And OpenAI is telling investors to expect $200 billion of revenue by 2030, against a current quarter of $6.7 billion.

In 26 years of investing, and 40 years following markets and politics closely, we cannot recall a time when chief executives were free to make claims with so little evidence, so little logic, and so little pushback from the press. It would be funny if it were not so serious. Our view is unchanged: a bubble this large tends to end the way they always do, when technopia and finance collide and the whole thing stops at once.

Two threads, and both point at the same shift: the age of frictionless globalisation is closing, and the bill is landing in markets.

Immigration keeps moving to the centre of the story. Many Australians want the total numbers cut, the government has seen the rise of One Nation and decided it needs to act, and the whole question is tied tightly to the property market. Alan Kohler has laid out how migration settings and construction reform combined to build the housing squeeze in the first place.

Step back and the deeper issue is concentration. The Australian economy has always leaned on a lop-sided mix, heavy on residential property for growth and on external markets like China. Nowhere is that clearer than in the index most people think of as diversified.

We have made the case before for rare earths and the wider critical-metals theme. Here is the glaring problem that will most likely slow any Western supply chain: the West has largely forgotten how to turn the raw metal into a usable product.

This is where China holds a serious and deliberate lead. Processing know-how is the choke point, not the ore, and it is not knowledge China will hand over. Beijing is now restricting the movement of the very experts who hold it. One report captures the shortage perfectly: the talent scramble has grown so acute that an 86-year-old specialist has been lured back out of retirement, because so much of the industry's expertise has migrated to China.

We do not think it is widely enough understood how deep this gap runs, across rare earths but also advanced and even basic engineering. You cannot conjure that expertise overnight. Building it back will take years of intense, focused effort before the West can stand up the infrastructure it now says it wants.

This week's podcast and the full show notes accompany this edition. Steve, Tom and Jacob work through the AI bubble, the two-bet index and where the real risks sit from here.

We will keep updating the sector scorecards, momentum indicators and macro notes as these themes unfold. If conditions shift, you will see it in the Wells calls, the Signals and Noise Premium updates and the portfolio insights.