Jean-Claude

📚 PDF ⏳ 7 min 📖 6

Remember when

1998 was not the same. The Bank of England had just been given independence in 1997, Tony Blair had just won his first landslide election and the first five years of “New Labour”, as they called themselves, were quite sensible. 

The Bank had one opportunity to demonstrate their credentials regarding inflation and they took it. The base rate itself was high at 7.5%, so a 30 year rate of 5.2% was not a surprise. I was there and London was absolutely booming in the early Blair years, not that we knew it or acknowledged it. It was expected. 

Contrast that with now. The 30 year rate is basically the same, just over 5.5% but the UK Base rate is now 3.75%. The market has zero confidence, they are bidding rates upward. The economy isn’t moving either, it's probably shrinking. London’s youth unemployment rate is 24% which is distinctly Mediterranean in its characteristic. 

The incumbent government has until 2029 to turn it around, but things do not look good. They must spend an awful lot less, perhaps a 20% cut in government spending is needed. It won’t happen and they will borrow an awful lot of money between now and then ....and history will repeat itself.  

A £10,000 investment in the iShares UK 15 year + Bond index has yielded -43% over 5 years (£2 billion in AUM). These are supposed to be the most secure investments that money can buy. They are exactly the asset classes that UK pension funds pile into to “protect investors”. It has been catastrophically value destructive. The factsheet is here. 

It’s an extraordinary performance really from an asset that has no asymmetry, the best you can hope for is to collect the interest and get your money back. 

The view from the top?

More negativity this week following the increase in interest rates in Australia. It’s understandable too given the entire economy seems to stand on the pin head of house prices. Happily, house prices do not go down in Australia because gravity doesn’t work here in the same way it does in the northern hemisphere. 

I do wonder though. I bumped into a local estate agent near us on the weekend. He was (is) very handsome, wears a $10,000 suit, a Rolex watch and drives a quarter of a million dollar car. He was drinking an implausibly small coffee too, an oat milk thimblechino perhaps. I look for signs, and the universe sent me one with a fake tan and a Cucinelli suit. 

Elsewhere, China’s property index has given up all the gains since 2006 in the last four years. Looking more closely, the index has gone from 111 to 87. A 21% fall, it’s not great but it’s hardly the end of the world. It can happen it seems, even in property obsessed countries. 

Interest rates aren’t going to be the only blow to property this month, next week's budget won’t help much either if the negative gearing changes go through. I imagine there will be panic if Australian house prices were to drop 30%. 

It makes me glad I am somewhat inured against holding assets that have a habit of doing exactly that. 

Is it true?

I confess to very much liking Taleb’s books and very much not liking him. A hidden asymmetry perhaps. Increasingly though, much like his guidance, I look at investments a bit differently now with the lens of “is this true”. In my own opinion, too much of what goes on, is pretend. A few examples:

From 2023-2025 of the 800,000 jobs created in Australia, 80% of them were in the government funded sector. It’s not really possible in a country as small as this one. It simply is not true then that the Australian economy is where it was a few years ago. Note of course that a person who gets paid $100,000 to photocopy pieces of paper and send them to another department adds the same to GDP as someone who gets $100,000 to fix electricity poles or perform research into medical cures. They are not the same, some of the efforts compound into the future, the others do not. 

Obviously then, productivity has collapsed. I know it's harsh, and the public servants amongst you who email me about how hard you used to work. I believe you. I just do not believe that the newly recruited 800,000 people are the same. 

Specifically then, to what I believe is true. 

Some countries are pursuing energy policies that deny this particular truth. The United Kingdom, Germany, Australia. Their policy settings are such that they imagine cutting their way to wealth. Dimmer light bulbs, no air-conditioning, taxes on flights, taxes on roads, housing policy that mandates insulation in every orifice of a home. In short, the politics of less. They no longer drill for oil, they stopped or have banned nuclear and as a consequence their energy prices and general price levels are now soaring. We must also acknowledge that it has worked. Their energy consumption has dropped, just like their relative standard of living. 

Why, though? Because they were told something that was not true. Specifically, that renewables are cheaper and sufficient. Of course optically, it is true because solar is free but it is the last joule of energy that sets the price and for most of the time it cannot be renewables. 

There is overwhelming denial of this point, indeed many of you will not agree, which is great. The point is, it creates opportunity because the things that are true (in my view) are ignored and are cheaper than they should be, while the fraudulent political marketing campaigns absorb more capital than they should prior to their eventual collapse. 

A 1 million x increase in computing demand. I also believe this is true. Indeed so much so that I ran my own calculation, which was ridiculously simple but I wanted to check. 

I use a 2019 Mac 2.3 GHz 8-Core Intel Core i9 (the last Intel Mac; it was a mistake, should have waited for the M Series)

In 2023 that was pretty much all I used. CPU at roughly 30%, say 8 hours per day.  

In 2026

2019 Mac 2.3 GHz 8-Core Intel Core i9. CPU usage same. 

2 Linux boxes (Hetzner €8/month excellent value - runs my Clawd and Hermes bots)

Codex Pro (4 hours/day at max)

Claude Pro (4 hours/day at max)

I threw it into various AI toys, estimates varied from a 300x increase to a 50,000x increase in my computing demand. Not 1 million for me then but several orders of magnitude increase. That is perhaps less interesting than the fact that I could easily consume 100x more than that today if the machines were just faster at doing what I ask. If they were it would generate even more demand from me and so on.

Essentially because of AI, there is unlimited demand for compute. If we can now build intelligence through energy consumption there must be unlimited demand for energy too. Perhaps there always was, but never was it so easily convertible into something so malleable. That is the crux of the issue. 

Energy always was the economy, but now more than ever because it is so easy to convert it now into useful work. 

Jean-Claude

This graphic is brilliant. Under the banner of “if Europe had made AI”. It’s meant as a joke, but everything on there is real. Every regulation, every pop up, every comment. They all exist in Europe. Apologies it is a little bit hard to read unless you zoom in. So, my favourites:

The sub-heading: Sovereignty, Conformity and Hesitation: Beta Phase III. 

Response blocked pending ethics committee review. 

You have the right to obtain a copy of your prompts

3 of 5 daily prompts remaining

We share your prompts with 412 regulatory agencies

Uptime 94%

 I would think there are 100 jokes on that one page. All real.

Euro-Trash

For most people, Le Mistral is an unpleasant cold wind that blows through Southern France. It would not have been my first choice of name for the leading European AI model. Mistral Medium 3.5, the latest update, was released last week.

Mistral has a context window of 128k, the smallest available of the open source international models. It ranks lowest amongst all of them for performance and yet somehow manages to be the most expensive. 17x more than DeepSeek V4 Flash (which has 1 million context tokens). 

Overall it ranks number 10 amongst the open weight models. Notably this does not include any closed source models that we are all familiar with GPT, Claude, Gemini etc. I can’t really think of any reason why someone would choose to use it given its inferiority and expense. 

It’s basically crapola and probably takes the whole of August off too. 

It’s not that Europeans can’t work out how to do it, they have. They do not have the data centres, the data, the chips or the energy to power them. So when they train a model, it is slow and out of date by the time it arrives. This trend is not going to slow down and there will be a massive reckoning for Europe at some point when they fall so far behind that it can no longer be tolerated. 

Whether any of us like it or not, we are basically just using API keys now to send money to America. 

I did a personal inventory believing I use precisely zero European products. Turned out not to be true:

Finish: Dishwasher tablets

Nurofen: Occasionally over-indulge

Spotify: music subscription

Chips: iPhone chip schema designed by ARM Holdings UK

Hardly overwhelming though. 

Further information

Our April 2026 report to investors can be found here.