S2N spotlight

Over the last few months I have become increasingly bearish. You would have picked it up in my letters, but I have never shouted it out. I have been gently nudging towards a portfolio focused on defence.

Well, today I am hoarse, as I can no longer subdue my sense of foreboding that we are on the cusp of something big, and yes, I AM SHOUTING.

I think there is a very real chance that the major equity markets have already topped and we are staring down the abyss of a major correction.

The obvious question is why I am now Dr Bearman.

To attempt to answer this, we need to take a look at the decoy. The reason why every bull is still forecasting higher markets.

The evidence is pretty solid. Its consensus and the reason why so few are prepared to put their proverbial balls on the line. Pardon the expression, but I don’t have a unisex equivalent, and remember I said I am shouting, so let us forget about manners.

Take a look at the S&P 500 forward earnings growth expectations. We are expecting another 12.46% growth against a 100-year average of 5.60%. That is according to Standard and Poor's consensus data.

Naturally a linear chart like this enhances the compounding effect, which is why I wanted to lead with it because this is what most people see and base their decisions on.

Now let’s remove some of the visual bias.

On a log scale, the story is less dramatic, but the conclusion does not change.

I have written in the past that it is very difficul, near impossible to sustain corporate earnings growth above GDP growth without productivity gains filling the gap. There is no doubt productivity will help the earnings picture.

What I am sharing, as I said, is the decoy; this is what everyone is focused on. Like drinking from Jensen Huang’s Kool aid when he said yesterday that Nvidia will earn $1 trillion in sales by 2027.

My point:

What I have noticed is that the market and big business are asleep at the wheel with regards the coming job layoffs.

Most people are aware that there is an AI revolution taking place but most are fooling themselves (in psychology it's called a defence mechanism) that there is far less need for employing people to do repetitive work. You will be amazed at how much repetition there is in an ordinary job.

Sadly the lower down in the corporate food chain you are the easier it is to replace you.

I am not saying that AI replaces all human input and I am not saying we won't invent new industries in the future. But that is the more distant future. Before we get there, it is going to be disruptive and harsh.

Another observation for those who still think they are so smart that a machine cannot do their job because it lacks experience let me say this.

Dario Amodei has been speaking about “the exponential” in AI advancement for more than 5 years if you follow his work. The thesis is that by throwing more compute and training, models get predictably smarter.

In a recent podcast Dario said that we are coming towards the end of the exponential. Many people misunderstood this to mean that models are going to stop getting as smart. That is not what he meant he was saying that we are close to the point where AI Models are going to be smarter than humans in every endeavour they are focused on.

That is not even focusing on the growth in the robotics space. You should read this excellent article how Travis Kalanick the co-founder of Uber has bet big on this.

S2N observations

Very interesting to see how internet traffic has collapsed on many websites. Now people just use ChatGPT or the other LLMs.

This might drive home the point where the focus is with regards economic growth. One thing is for sure most AI companies and the investors in the data centres have no idea how they are going to pay off the debt used to finance all the investment.

Sam Altman's favourite line when getting push back how they can justify the debt and earnings projections is that if you don't like OpenAi there are other investors who do believe.

A quick and visually brief lesson on tokens. You should get used to this term as this is the air AI breathes.

Tokens are usually more than a letter but less than a word. Some words like cat are 1 token. When an AI model gets a question it references it with its short term memory, this is called “context”.

What blew me away and anyone heavily using AI is that last month the more recent models were launched with a context window of 1 million tokens.

Just in case the markets didn't have a potential jobs crisis and a fiscal debt cliff to navigate, we have the oil complex and major wars in session.

S2N screener alert

I ran out of time and I am typing on my mobile.

S2N performance review

S2N Chart Gallery