A data center stock worth considering

A torque play on the AI and tariff trade

Three weeks ago on youtube I made a video where I discussed the delta between the pre deep seek data center trade and post deep seek data center trade. Its linked below, but I think the first 10 minutes are worth watching because while it doesn’t directly address multiple expansion and contraction it explains why many data center names have been on massive downtrends since the end of January.

That was before Tuesday afternoon, when Nvidia announced that because of the evil US Government intervening with their exports to China, they would have to take a sizable hit to top line revenue. At least 8%. Pay no mind to the fact that again, business was starting to slow due to lower expected CAPEX spend on data center growth by the hyper scalers and that the expected H2 2025 and 2026 growth rates weren’t going to be nearly as bullish as the stock was pricing in no matter what any government did….no it’s all the US governments fault. They are to blame for the 13% drawdown.

Outside of the utilities, I’ve been pretty bearish on the data center trade since mid las year. In July I published a piece called ‘AI is dead, long live tech’, in which I said the big risk to AI/tech stocks was the dollar blow up and the fact that the latter half of 2025 and 2026 might not be as bullish as a $140+ Nvidia stock price/$200AVGO/$160 VRT/etc etc was pricing in.

Both of these fears turned out to be correct. With the dollar getting smashed since the start of the Trump administration leading to less foreign ownership of stocks, almost entirely in the tech (AI) sector, and DeepSeek lowering the estimated demand needs of most semi conductors and data center products over the next 18 months.

Now, nine months later, I think we’ve started seeing enough of a washout in some names to make them worth at least looking at. Micron, which I wrote up a few weeks ago and talked about in the above video has proven resilient over the last week of pain in the semi space, so far holding the key level of 68 as Nvidia blew up.

That doesn’t mean I’m bullish on the sector. Nvidia has been net bought by retail almost every day since their last earnings report. Even with the massive pullback on Wednesday and Thursday, there is little fear in that name. A similar story is playing out with Broadcom. The semi conductor bulls keep pointing toward earnings estimates that are only now just starting to decline and saying ‘look this is the cheapest these stocks have been since pre covid’. Which means they still have a lot more potential downside if those estimates start pulling back. It’s the opposite of a sector specific buyable set up, as if the estimates are right, the names that have held up might have 15 - 20% upside, but if they’re wrong, they need to price in the lower growth expectations, meaning the downside is potentially double those numbers. Heads I win a little, tails I lose a lot.

Not great as a long pitch for a cyclical industry.

This thesis might be wrong.

If we start to enter a recession the pressure on some C-levels to cut costs and operate with a thinner labor force AI adoption, whether its the best or not, or ready or not, might explode. This would make the E (earnings) for the bull case valuation rip higher, and with it the stock prices. (Note that utilities would end up being huge beneficiaries of this thesis as they would have sustained growth and energy demand for the next 15 years). Today’s company is interesting at current prices for two reasons. First, if my thesis is wrong it will be ready to explode higher. Second, even if my thesis is right its priced so inexpensively relative to its peers that it still has material upside and lower downside.