The Nuclear Bull Market is Here!

She’s so fine, there’s no telling where the money went… Simply Irresistible, Robert Palmer

I was a greenhorn in late 1999 when MarketWatch TV asked me if I’d come down to whichever local Baltimore network affiliate aired the show to talk about a stock report I’d written. I wore a suit and everything.

You may not remember but during the height of the internet bubble, CNBC was not the only stock market show in town. Internet stocks were a family dinner topic in some households – of course the networks got in on it…

I’m pretty sure the stock MarketWatch wanted to talk about was called Barcode.com (of course it was a dotcom!) – I might have the name wrong, that was a long time ago…

Barcode.com (if that’s really your name) was a mobile phone software company. The most promising offering was a program that let you use your phone to scan barcodes and comparison shops.

Not revolutionary, really just an incremental step in making the internet useful. I have no idea how long the company managed to tread water before finally going under. It lasted probably about as long as Marketwatch TV.

But the idea was perfectly sound. People use their phones to scan QR codes all the time. The Netherlands’ NXP Semiconductor (NASDAQ: NXPI) pioneered near-field communication (NFC) chips, which cell phones can read and access information. It’s a $60 billion dollar company.

Something I pointed out during that interview – Barcode.com was way ahead of its time. Cell phones weren’t really sophisticated enough, and nobody was clamoring for that scanning service.

There were a lot of internet companies like that. Plenty of sound ideas, plenty of bad timing. And way too much money fueling the internet stock party. There’s still no telling where all the money went…fancy office spaces and corporate junkets.

I expect it’s pretty obvious where I’m going with this…

The AI Bubble

As much as I enjoy the debate, those making the argument that AI is not a bubble are wrong. Of course it’s a bubble! And bubbles are great for making loot, provided you get in ahead of the crowd.

Bubbles are how markets work…

Innovation/opportunity attracts investment and the assets involved rise in price until it is clear that there is over-investment, capital is being allocated poorly, then the asset prices fall.

For AI, like every other investment trend, the only question is when, not if.

Snowflake (NASDAQ: SNOW) was one of the early AI movers. It ran from $135 a share to $233 from mid-2023 to early 2024. It’s since given most of that gain back after earnings growth started to slow.

A year ago, Snowflake was thrilled to have placed an order for 1,000 of Nvidia’s H100 graphic processing units (GPUs) – the backbone of AI. I’m sure Nvidia was pretty pleased too, H100s cost $25,000 each.

But during a recent conference presentation, SNOW Chief Financial Officer Mike Scarpelli said: “I’m not going to buy any more GPUs until I see the revenue to support it.”

Pretty clear Scarpelli thinks capital spent on GPUs is being poorly allocated right now.

I’ve shared this image with you before, but it bears repeating:

NVDA rev

Nvidia is expected to finish this fiscal year with $120 billion in revenue. Microsoft alone will account for $24 billion. Microsoft’s CAPEX is around $40 billion for 2024. Back in January, Meta said it was in the process of buying 350,000 H100 GPUs.

There will come a time when these big spenders lay down their own version of Scarpelli’s Law of GPU Buying.

I can’t tell you when that day will come. My experience tells me it won’t be much longer – first quarter or two of 2025. But I also know that markets can remain irrational for a long time…

One thing I can say for sure: when one MAG7 company scales back its spending, they all will. When that happens, just get out of the way, especially with chip stocks.

Nuclear Percolator

Now, Hammer and I have been talking about nuclear power stocks since April. I’m not gonna share every article here today, just take a scroll through the Outsider Club website and you’ll see ‘em.

We tend to call nuclear stocks “picks and shovels” plays on AI because of the electricity demands coming from all these AI data centers. The U.S. has literally never seen electricity demand grow this much, this fast. Nuclear has burst back onto the scene as a top option for new electricity generation. Right now, the world’s biggest banks and investment funds are in New York, pledging to triple global nuclear power output by 2030…

The irony of the nuclear “picks and shovels” trade is that it too will become a bubble. It already is…

I know, the words “picks and shovels” might sound conservative, defensive, or even stable. The big utilities like Southern (NYSE: SO) and Dominion (NYSE: D) probably will be pretty stable investments…

But nuclear will follow the bubble playbook. Which means you can still make some loot.

The smaller companies developing the nifty small modular and microreactors (NuScale, Oklo and Nano Nuclear) have already made triple-digit runs. There’s more to come – these stocks are still on the fringe of most investors’ awareness.

Same thing for uranium miners. There’s going to be a mad rush to open new uranium reserves since so much of our supply is dependent on stupid Russia. We could be trading penny uranium stocks soon…oh boy!

The only U.S. uranium refiner is Centrus Energy (NYSE: LEU). I wrote it up back in June and mentioned it again Friday, after the Constellation/Microsoft nuclear power deal.

Centrus was up +10% yesterday and another 5% today. It is valued below $1 billion, it trades at just about 2X revenue, it has $50 million net cash and a trailing P/E of 9. Strangely, next year’s earnings are expected to fall, giving the company a forward P/E of 17. Maybe not so great. However, given the speed with which the nuclear story in the U.S. is changing, I would not be surprised at all if forward estimates for Centrus started going higher.

Cheers,

Briton Ryle
Chief Investment Strategist
Outsider Club

X/Twitter: https://twitter.com/BritonRyle

The Nuclear Bull Market is Here!

Latest Articles