Preparing for the Bust

Nvidia has been the main beneficiary of the AI Cycle that started in late 2022. The launch of Chat GPT was the inception of that whole process.

The voracious appetite for compute — either to train Models or to power them (i.e. Inference) turned NVDA into a $5 trillion company.

Now NVDA is already ~20% below its peak – and preparing for another leg down.

The issue however is not as simple as that. While NVDA was the main beneficiary of the Cycle – it also became the Cycle!

I tactically shorted NVDA a number of times over the past 2 years, managing to net significant gains from sharp price reversals. Arguably I should have looked for areas to go long and chill, but that’s a discussion for another time.

In fact, I almost bought the Tariff Tantrum dip at $100ish, as a brink and back trade. But I didn’t and I let my bearishness get the best of me!

Now the stars are aligning – and the last pieces on Philoinvestor build up to this trade.

• Layoffs-as-a-Service: How mass layoffs under the banner of AI Productivity creates the reversal of this Boom/Bust process.
• AI-ran: The energy chokepoint that will result in an AI chokepoint. We made this post open to the public.
• Trump is Pot Committed: A game theory take on the geopolitical setup around the Persian Gulf and who might not make it out alive.

The setup is that the war in and around Iran is still starting, that the AI Capex Cycle is nearing its final stages and that this last stage entails mass layoffs to prolong the Capex cycle just a little bit longer.

News flow already signals that META is looking to lay off 20% of its workforce, in an attempt to signal real AI productivity and save cash to invest more in AI 🤦🏻‍♂️

Now, you might ask.

Philo, if the Capex cycle is still going and there will be ~1 million in staffers fired to cover up for that cash deficit....

Why are you short?

I am already short a basket of AI stocks, incepted in a Philochat from a few weeks ago with Google only making the cut on Friday.

Here are the points to my thesis, one by one.