By: Wilson
I came across some charts today from The Crude Chronicles that motivated me to write this piece. It helps that Amazon surprised everyone today that it is going to spend $200 billion in capex.
Source: The Crude Chronicles
That’s an eye-opening chart.
The Amazon announcement will make it 2 tech companies spending more money on the AI buildout than the entire “Western” oil & gas companies.
And who said oil & gas is a capex-intensive business?
In a November Memo I wrote titled, “Shitty Businesses.” People who invest in tech aren’t familiar with what a shitty business really is. In oil and gas, that’s a dime a dozen.
Why? Because you need to spend money to make money.
What’s the holy grail of running a business? A business that can make money without you having to invest/spend money.
Hedge funds are great businesses. Your operating costs are fixed, performance fees offer optionality upside, and scale begets scale. But the problem with hedge funds is that you need 1) a track record, 2) the ability to deliver market-beating returns, and 3) size.
So the barriers to entry are high because you can’t just start one tomorrow (well, you can, but you get the point).
Let’s use another example, the Substack subscription business.
Your product is your research/article, you spend time, publish it, and your economies of scale are infinite. One article could get you 1, 100, or 1,000 subscribers. The incremental cost of getting an additional subscriber is zero.
Over time, if you maintain a good reputation with a good track record, scale begets scale, readership begets readership, and the virtuous cycle grows. The key to maintaining a profitable/growing research service is to 1) control your costs, 2) control your overhead, and 3) improve the quality of the research/product. The issue is getting the publicity needed and the initial readership. That process is a grind, and it’s one of the many reasons why readership eventually funnels upward (to the most popular services).
Now, take the subscription business and add a new element to it. Instead of original content generation, you “spend” money to create content. You hire people to write articles. Without an effective way of controlling this cost, the returns are extremely unpredictable.
Sure, you might scale faster by branching out into more sectors or topics, but you also run the risk of diminishing the economies of scale of the business. In addition, you introduce key man risk where one of your contributors might decide to jump ship one day. In essence, the returns greatly diminish when you introduce this new element (hiring) into the equation. You no longer enjoy the infinite economies of scale.
What does that have to do with tech capex spend?
Tech was always considered a low-capex intensity business. That’s what made it so attractive for investors. You spend X on developing software or a product, you sell it, and the incremental cost of selling additional units were diminimis.
In the case of Meta, you have a social media platform like Instagram; scale begets scale, content begets content, and advertisers rush to your platform to spend money. You control user data, you know what time of the day they log on, what they see, what they like, and you can tailor each advertisement to their usage/behavior. You charge a premium advertising rate relative to peers because you have a competitive advantage. So long as people use the service, you have an enduring competitive advantage.
And the same scale dominance applies to every other major tech company, Amazon, Alphabet, you name it. As time goes on, you get bigger and bigger, and the smaller competitors exit the business. That’s how you build shareholder value.
But AI has upended all of that. If you don’t spend money on developing the infrastructure or the AI models, you risk falling behind, and if you fall behind, you are dead. Or at least that’s how a lot of the big tech guys are thinking about this. Apple is an exception here, and what’s ironic is that its failure in rushing towards a competitive AI model might actually turn out to be a benefit later on.
Only time will tell, but for how much longer will investors endure this period of rapid capex expansion?
And this is best illustrated by another amazing chart from The Crude Chronicles.
What preceded the crazy capex spend we saw from 2004 to 2014 in oil?
The decade-plus oil bear market we had to go through from 2015 to today.
What does that mean for tech? I don’t know, but this level of spending can’t continue.
Parting Words
Energy investors learned this the hard way. Fracking was revolutionary, but most investors didn’t make money from it. Oil and gas companies borrowed money to invest, leaving investors holding the bag. Growing production was the name of the game; whoever had the highest growth was winning all the investor attention. Then it all ended in 2014, and again in 2018, and again in 2020. Now discipline is the name of the game; everyone focuses on delivering free cash flow for investors, paying down debt, and maintaining rock-solid balance sheets.
Energy companies didn’t have this “come to Jesus” moment overnight. The market pounded this mentality into them for 5 years AFTER oil prices collapsed. For god sake, from 2017 to 2018, US shale grew US crude oil production by over ~150k b/d per month (+1.8 million b/d per year).
And if you go back in time, no one in energy thought, “Well, shit, if I grow this much and my competitor grows this much, doesn’t that mean we are all screwed?” Yes, but where was the discipline when it was obvious?
I do wonder if history rhymes here for the tech boys chasing the dream of AI. I’m not a specialist in that field, so my opinion on this shouldn’t matter. But I’ve seen a thing or two about bear markets and technological innovations; sometimes, they aren’t necessarily good for investor pocketbooks.
Analyst’s Disclosure: I/we have no stock, option or similar derivative position in any of the companies mentioned, and no plans to initiate any such positions within the next 72 hours.