One week / one topic: AI or die

What happened?

On Thursday, Nvidia announced yet another set of mind-blowingly strong quarterly results.

And yet – after initially climbing 5% – the stock staged an 8% reversal and closed -3.2% down for the day, dragging down global equities.

Bear in mind, we are talking about market cap swings to the tune of hundreds of billions on one single stock here…

Meanwhile, in the background the AI race is actually accelerating, it that’s even possible…:

• Alphabet released Gemini-3, a new top-performing model that rekindled ‘winner-takes-all’ concerns

• Brookfield launched a $100bn fund to ‘build the backbone of AI’

• Chinese hyperscalers continue to (more or less) keep up with US behemoths, despite spending much less (remember DeepSeek?)

And that’s just a partial recap of one week’s worth of developments!

At the same time, concerns about how these capital expenditures are funded are becoming more and more prominent.

We all know that the US hyperscalers are wonderful businesses with superb profitability and huge moats, that is not the question here.

The increasing scrutiny is more focused on the gradual, but quick, change in their capital allocation choices:

1. We are making so much money that we are just going to return a lot it to shareholders in the form of stock buybacks

2. Wait, this AI thing has immense potential… let’s spend some of our free cash flow to finance this

3. Oh no, we need to keep up with competitors! Pedal to the metal then, let’s ship even more of our free cash flow to Jensen and friends 😊

4. Hold my beer – let’s issue some debt to fund this!

Lo and behold, Oracle’s credit default swaps (a measure of the riskiness of its debt) have been getting a lot of attention lately as investors look for hedges – if not for ways to take the other side of the ‘AI trade’ outright.

While all the sound and fury about AI might feel unjustified or overdone on some days, I believe that the theme is so pervasive that you just can’t ignore it.

It not only has real-time, very tangible effects on capital investments, equity index returns and trade sanctions. The advent of AI is also raising deeper, longer-term questions around the future of jobs, education and productivity.

What should we do then?

Zero in on the ‘AI or die’ phenomenon, or take a step back to evaluate the bigger picture?

Our observations

• Fundamentals: The latest labor data was weak-ish, while Fedspeak still sounds skeptical about further rate cuts. No surprise concerns about a policy mistake are starting to emerge…

• Price action: US retail investors represent a large and growing portion of trading volumes, and in the recent past they have (successfully) bought any dips in performance. Will they come the rescue once more, or perhaps institutions can pick up the slack between now and year-end? What about neither?

• Investor beliefs: As more and more people use AI-based technologies, there is wide consensus that these tools are already changing the way we live. Meanwhile, initial investors enthusiasm might have given way to more nuanced considerations: AGI or ROI? Winner-takes-all or oligopoly? Fee-based or open-source?

So what?

And so, here we are – once again dealing with radical uncertainty.

While no one knows what’s the final destination for the AI race – unless AGI is already here and is just toying with us for the fun of it, which unfortunately I can’t 100% rule out – there are some things we can do.

If information is like food and it has a shelf life (h/t JP Rangaswami), we should then be careful about what we ingest and give the most consideration to information which has the longest shelf life.

Through that lens, paying too much attention to daily blow-by-blow commentary is unlikely to be good for you.

In the hope that ‘poetic justice’ market developments will ultimately reward restraint and delayed gratification, we like our inflation-linked bonds positions even more.

For starters, compared to nominal government bonds, inflation risk is contractually removed. (Yes, I know CPI baskets are subject to change – but still)

Plus – even while allowing for the eminently credible concerns about inflation, or perhaps precisely because of them – you can point to the most attractive risk-free (…) real returns available in a very long time.

Maybe we should add to our positions?

Mood music: The Clash – Should I Stay or Should I Go?

By popular demand, here is the One week / One topic playlist