The Finance of AI

I wrote to you at the end of March (Q1), debunking some common investment myths. You should check out the article here. It's pretty interesting (if I do say so myself).

At the end of this quarter, I thought to write about something else that has captured our imaginations, AI.

If everyone does the right thing, it could be bad for everyone

I did British Parliamentary debate in university, and one common (smart but not so smart) argument we regularly use is "X thing is good, but if everyone did X thing, it'd be unsustainable"

For example, starting a business is good. But if everyone starts a business, who will be the employees of businesses? Or drinking water is good, but if we all drank water… beverage companies would go out of business.

You can make that argument in finance.

Frugality is good, but the global economy would collapse if everyone was frugal, made a budget, saved a large chunk of their income, and only spent on necessities and a few wants. Most (discretionary) businesses would go bankrupt if we were all fiscally responsible.

Now, enter AI and other technological advancements.

Disclaimer: Everything you're about to read is a thought experiment.

When looking at business case studies (and probably in business school), we've heard several stories of businesses that ignored the next wave of advancement only to get completely disrupted. We all know popular examples of companies that refused to change with the times and ended up dying out.

Think newspaper companies vs internet companies. Or physical mail companies vs email. Or cinemas (or theatres) and cable vs streaming. Or… you get the idea.

We can all name several industries that, because the old guard refused to get into the new thing, the business fell out of popularity and ultimately either went bankrupt or was sold for scraps.

Nobody wants to be, work for, or be the CEO of that company.

So, everyone's doing something different.

Enter AI

Over the last 12-18 months, every business or person (who isn't living under a rock) has either incorporated AI into their daily lives or is actively thinking, spending time and money to figure out how they can incorporate AI into their workflow.

The largest tech companies in the world are no exception.

Capital expenditure (capex) at some big tech companies is rising fast, mostly due to the AI arms race.

John Huber makes 3 key observations from this higher capex spending

• Higher capex is not a bad thing if the returns on the investment are good. You know the saying, "You spend money to make money". The only issue here is that it’s unclear what the returns on these significant AI investments will be. We're all tinkering with AI, but there aren't clear ways for these companies to make more money from these investments. Not as of today, at least.
• Growth in invested capital at these businesses is accelerating quickly, and all else equal, which means lower returns on invested capital and lower earnings growth in the future. Because more firms will share the additional returns, everyone will get a smaller piece of this imaginary pie.

Think of it like having a bachelor's degree in the 1900s. Back then, having a degree was the guaranteed ticket to a middle-class life. Now, a bachelor's degree is the "bare minimum". No one gets "specially rewarded" for a bachelor's. The "excess" returns (or profit) from having a bachelor's degree have been competed away.

• Rising capex will hit profits eventually in the form of higher depreciation: The way accounting works; this higher capex will not hit the profits today. They will come through the income statement over time as depreciation. If revenue doesn't increase at a higher rate than the spending, then profit margins may be worse in the future.

Perhaps one reason why early adopters of new technologies (in the past) ended up making more profits was because others did not adopt the technology as early.

If everyone is an "early adopter", most of the additional profit to be made will be shared equally. In other words, everyone ends up spending more money to make the same profits they were making before.

But remember, in the gold rush, the best person to be is the shovel seller. So perhaps those selling the tools (chips?) needed to tap gold (AI) are the winners in all of this? While the rest of us spend more time, energy, and money to make the same money we were making prior.

If everyone does AI, the excess profits may be competed away, so no one benefits from the significant efficiency boost after all.

If we all learn from one mistake, we could be making a different mistake

If everyone does the right thing, it could be bad for everyone

Final Thoughts

Did you know that crypto and AI are deeply related? The crypto boom (and bust) of 2020-2022 may have ushered in the AI boom from 2022 to date.

The connection is chips.

If you want to mine Bitcoin, you need a lot of Graphics Processing Units (GPUs), the specialized computer chips that were previously mainly used by gamers.

And if you want to build a large language model (LLM) (like ChatGPT), you need a lot of the same GPUs.

The crypto boom created a huge demand for GPUs, which led GPU makers (mainly Nvidia) to make more GPUs.

But then crypto winter happened, and miners didn't need all these extra GPUs.

Then, companies that wanted to build LLMs could easily acquire lots of GPUs to train their models. That led to a transformative improvement in the power of those models.

And now we are in an AI boom, with enormous demand for those chips. But crypto got us there.

How do you strike gold twice? Or rather, sell the shovels for 2 of the biggest gold rushes in history? Be Nvidia