2-bit Tournament

Imagine you go to the park to play a game—soccer. You bring a ball, maybe. A soccer game only needs one ball. There's probably already one there. That's it.

You show up at the park and instead of soccer, the game is jousting. You have no horse, no lance, no suit of armor, no gauntlets, no helmet, maybe some puny soccer shin guards, but what good will those do?

This is a bit how software investors feel as the world pivots to AI. For two decades, they've been able to rely on a high-margin/low-capital lens to evaluate companies. Now those companies are asking them to think about real "investment."

Once upon a time

"Investment" used to mean putting money upfront for a machine, a factory, some land and getting that money back later. If it was a smart investment, you'd even get a return.

"Investment" today mostly means buying financial stuff—stocks, bonds, triple-levered crypto ETFs. It also still means investing in real stuff—machines, buildings, and things.

Maybe you want to make a camera. You'd need to design the camera. You need a factory. You'd need the "camera making machine" whatever that is. You'd need to get these things before you make a single camera. It would be the same for one camera or ten or a hundred. You'd need to spend "capital" upfront to produce the thing you want to sell.

From the start of the industrial revolution through the end of the nineteen hundreds, this was how most companies worked.

Management magic

Then, one day, some punchy manager woke up and thought "hm, I wonder if I could pay for investments as I go?" The software era made this possible. Companies could plug in coders, product managers and account executives as they needed them. These weren't upfront costs. They weren't accounting "investments." Companies just paid salaries. Better still, they could pay employees with stock. Tech companies paid staff in shares as a backdoor way of raising capital and funded the business as it went along.

Once engineers produced the software, it didn't cost much to send 1 or ten or ten million products—it was just lines of code. The software that engineers wrote paid back over time, like an investment. Compensation expenses ran through companies' profit and loss statement, like other staff—accountants, janitors, execs. Accounting failed to keep up. These costs didn't go on the balance sheet like your "camera making machine."

Companies were happy as software ate the world. Businesses did well. Coders were happy as the company's share prices rose and the value of their compensation soared.

All of a sudden, ChatGPT marched into the scene and threw a wrench in the machine. It even started to code.

CEOs at the biggest tech companies had a range of reactions--I'm guessing:

Satya: "Shit"

Zuck: "Shit"

Sundar: "Shit"

Cook: "Shit"

Jensen: "Let's go!"

The original ChatGPT tipped big software companies into an arms race. AI would, maybe, take over everything. The companies needed to win.

But AI is different than software. Companies (mostly) can't pay for a server using equity. GPUs, networking kit, memory—these are investments. Companies need to spend a lot (A LOT!) upfront to make it work. They need to spend before any income comes in. And once it works, it’s not like shipping the same software update to every customer. Each time someone uses AI, the answer is different. AI uses resources—energy, logic—to deliver that. Each use has an expense.

Hm. Upfront costs. Real cost of production. This sounds a lot like the pre-software era. It's not the first time software has crossed over into the real world, though.

When Amazon plowed billions into warehouses, forklifts, planes and trucks to build out its logistics investors asked "wtf?" Sometimes they punished its shares. Now, connecting the dots in hindsight, it's obvious that this was a clever move. But a lot of people didn't think so at the time.

A new game

Amazon is now frantically spending on AI. So is Meta and Microsoft and ChatGPT and so is Google. Lamplighter pokes fun at Google from time to time. Investors got a little skittish after it said it would spend $75 billion on capital expenditures—computing gear—in 2025. SEVENTY-FIVE BILLION! Double what it spent in 2024, which was already hot from AI spending. Triple what it spent in 2021.

Rather than throwing up all over this plan, though, you can see it as a return to a more traditional business model, one with upfront costs and returns over time. You can see this start to show up in its P&L. R&D expenses fell from 15% of revenue in 2020 to 14% in 2024. Sales and marketing expenses fell from 10% in 2020 to 8% in 2024. Google employs fewer people now than it did in 2022. Spending is shifting from its P&L to its balance sheet. Management has said that already a quarter of the code written inside Google is done first by AI. And that's just internal use. Customers use it too.

Google’s also sending that pot of money towards data centers to support its Cloud business. Lack of racks held that business back last year. It still grew 30%—twice as fast as Google overall. Profits more than doubled. Cloud kept 45% of its growth as profits for the company. It’s a high-demand product with good visibility. It’s earned its share of the capital budget too.

Google has ridden a capital-curious model before. Its investment spending hit 15% of revenue last year. It had been 10-12% for the few years before that. In 2018, though it reached 18%. In 2014 it was 17%. 2025's $75 billion will be bigger both in dollar terms and relative to revenue than any of those years. In this new tournament, Google is spending to stay at the table of forefront AI models. The prize is bigger too. Google's been on a mission to organize the world's information. AI will help it and its users figure out what to do with it all.

Disclaimer: None of this is investment advice. It's meant to illustrate ways LCM thinks about investing. Things that LCM decides are good investments for LCM and its clients are based on many criteria, not all of which are covered here. Some or all of LCM's ideas may not be suitable for other investors. LCM does not recommend investing either long or short any position mentioned. LCM may own positions in some of the companies mentioned. Some of its ideas will lose money — investing entails risk. See full disclaimer here.