Giant AI spending has yet to dent this key metric
Add Axios as your preferred source to
see more of our stories on Google.
Big Tech's massive data center spending binge hasn't put too much of a dent in a core gauge of corporate investing prowess.
Why it matters: This number, return on invested capital — or ROIC, to balance sheet fanboys — aims to tell investors how good a company is at making money on all the cash that it has sunk into acquisitions, equipment and plants and other capital.
Zoom in: Profit margins don't incorporate the costs of investing in factories and equipment needed to create or sustain a company. For that you need to look at returns, percentages that tell you what a business makes once you take these expenses — and in the case of AI massive expenses — into account.
- Profit margins — which we did our best to dive into recently — can tell you about the basic economics of how much it costs to make a product, and how much you can sell it for.
- But profit margins don't incorporate the upfront cost of investing in the factories and equipment needed to create or sustain a company. (And in the case of AI data centers, these are massive.)
- To get a sense of that you need to look at returns, percentages that tell you how healthy earnings look, as a share of investment.
Zoom out: We're kind of obsessed with trying to figure out if anyone is generating actual returns on their massive AI investments, so we're looking at some slightly off-the-beaten-path metrics, like ROIC, to suss things out.
How it works: To calculate return on invested capital, you basically take after-tax operating profit at a company — a measure of the core business's earnings power — and divide it by average total invested capital, typically sourced from a company's quarterly balance sheet report.
The latest: Numbers produced by the most recent earnings cycle for companies like Amazon, Microsoft, Meta, Alphabet and Oracle show that so far, the giant investments these companies have made to build out their data center portfolios have done little damage to their ROIC (with the possible exception of Meta).
- Meta's ROIC did drop pretty sharply over the last year or so, falling to 22% from about 32% in the most recently reported quarter.
- Oracle — which reports a bit off the regular earnings schedule and is due to disclose results in September — has also seen a bit of a slide.
- On the other hand, big data center spenders like Alphabet and Amazon have seen a decent steady increase in ROIC in recent quarters.
- Microsoft has been more or less steady.
The big picture: This makes sense. Analysts have noted that companies that have other large healthy businesses, such as cloud computing, have been able to manage the expense of AI better than others.
Yes, but: That doesn't mean this giant collective bet on AI is costless. There's always the chance that these companies would be faring much better, financially speaking, if they weren't spending like drunken, AI-obsessed sailors.
Case in point: Check out the ROIC at Apple, which has soared in recent years after the company essentially decided not to participate in the manic capex expenditures its mega-cap tech peers have pursued.
Caveat: It's also worth noting that like any metric, ROIC has its flaws.
- For one thing, it may not fully capture the impact of big tech's off-balance sheet obligations, which are large — more than $1.5 trillion at least — and growing.
- If some share of that capital was fully factored in, it could make these numbers look quite a bit different.
