“Nothing makes you look older than attempting to look young.”
-Karl Lagerfeld
Two weeks ago, the world’s most significant economic indicator was revealed: Big Tech earnings. The results were remarkable. Microsoft and Amazon increased revenues by around 20%; Meta and Google, closer to 30%. For companies that already generate hundreds of billions of dollars a year, that level of growth is unheard of.
Beneath the surface, however, lurks some unsettling numbers. While Google’s cloud business nearly doubled, the company also spent $45 billion on capital expenditures in … one quarter. That’s nearly double what NASA spends in an entire year. Amazon spent even more: $54 billion. As a result of this extraordinary spending, both achieved what many deemed impossible: negative cash flows. In other words, despite the freakish amounts of cash they raked in, they somehow managed to burn more.
How? Answer: AI. Together, Amazon, Meta, Microsoft, and Google will spend roughly $1.5 trillion on data centers through next year. That’s roughly equivalent to the GDP of Indonesia. As with startups, however, the idea is that these crazy investments will eventually yield crazy returns. And, given the explosion in Big Tech’s earnings, some would argue they already are.
The more closely you look at those earnings, though, the uglier they get. While Google and Amazon more than tripled their net income, nearly 90% of that increase was attributable to mark-ups in their stakes in other AI companies. In fact, because they represent so much of the stock market, the earnings growth of the entire S&P 500 would be 40% lower if it weren’t for Amazon and Google’s AI investments. The numbers are telling us something important: Despite the illusion that their core business is on fire, most of the growth is attributable to future expectations surrounding the (unrealized) potential of AI.
Trillion-Dollar Questions
That’s why I was particularly interested to know how much actual AI revenue Big Tech made. To my annoyance (but not my surprise), they didn’t tell us. All we got was an “AI ARR” number from Amazon, an easily bullsh*ttable metric commonly used by early-stage startups who don’t want to tell you the real number (as it would disappoint you). As for the other tech companies, we got zilch — only hand-wavey, unverifiable anecdotes about how great AI is. What does that say about the strength of their existing AI business? Nothing good.
The question I was most eager for Big Tech to answer, however, was how much of their revenue was attributable to OpenAI and Anthropic. This is an immeasurably important question for two reasons:
Since OpenAI and Anthropic are both massively unprofitable, it’s unclear if they will actually survive in the long-run. This is a quietly acknowledged fact in pockets of the AI community, but it’s also starkly obvious when you look at the financials. In other words: If Big Tech is overly reliant on two vulnerable companies for revenue growth, that would imply their growth is unsustainable.
Since OpenAI and Anthropic are paying their bills purely with venture capital funding, and since Big Tech is responsible for roughly 40% of that funding, then that means any of the revenue derived from OpenAI and Anthropic by Big Tech, in reality, was derived from themselves. So, if Big Tech is too reliant on OpenAI and Anthropic, that would imply that Big Tech is propping up its own revenue growth with “recycled” money that isn’t actually real.
Now that you know why that question is so important, it might interest you to know whether Big Tech answered it. Answer: They did not. However, an excellent piece by Ed Zitron pointed to investment research that did. According to Ross Sandler of Barclays, OpenAI and Anthropic will make up 73% of Amazon’s AI revenue this year. In other words, if OpenAI and Anthropic went bust tomorrow, Amazon’s AI business would basically disappear.
After reading that statistic, my head was spinning. But then, a day later, I saw a number that’s arguably worse: According to Bloomberg, OpenAI was likely solely responsible for 70% of Microsoft’s AI sales last year. In other words, Big Tech’s reliance on OpenAI and Anthropic for growth is not only more significant than I originally thought, it’s also systemically pervasive.
What If It All Went Wrong?
How could stocks continue to climb despite this objectively concerning information? It’s not necessarily that investors don’t see it (though some probably don’t) — it’s that they’d rather not dwell on it. And you can understand why. Yes, it’d be bad if OpenAI imploded and brought Big Tech down with it — but on the other hand, it might be worse if you missed out on the gains of the largest technological revolution in history because you were worried about the finances of some AI startup. In other words: I don’t fault this market for hitting record highs.
It is worth pondering, however, what would happen if OpenAI did implode and this whole AI thing didn’t work out. In fact, that’s what I spent this week doing — and you’ll be glad to learn my conclusion is not that the stock market would collapse.
Two reasons: 1) While some valuations have gotten crazy (i.e. SpaceX), the most important ones (Big Tech) remain (somewhat) within the bounds of reality. Meta, Microsoft, and Amazon, for example, currently trade at multiples lower than before the AI trade really took off. That doesn’t mean they’re cheap — it just means they’re not completely insane. And 2) Big Tech was a perfectly sound business before AI existed. So if AI doesn’t work out, they’ll just go back to being juggernauts (albeit with significantly smaller balance sheets), but they won’t collapse.
So, what would happen then? The answer is more boring than you think. While certain pure-play AI companies would be shuttered or scooped, Big Tech valuations would simply get re-rated: Instead of trading at 30 times earnings, they’d trade at 20, as Wall Street stripped out expectations of future AI growth. That would be painful for investors, but not devastating.
For Big Tech, however, it would be catastrophic. Why? Because it would mean admitting the one thing they didn’t want to admit. It’s a simple truth about Big Tech — a truth they’ve spent countless years and hundreds of billions of dollars trying to cover up.
That truth is the following: They’re old.
Plastic Surgery
As much as Big Tech wants to see themselves as startups, the reality is they are mature companies whose best days are behind them. It’s been ten years since Apple came out with a successful new product (AirPods) and twelve since Meta completed an acquisition that mattered (WhatsApp). Since then, it’s been a painfully long (and expensive) game of throwing things at the wall and seeing what sticks. (For more on this, read another great piece by Zitron.)
Why don’t they just focus on their existing businesses? Why do they have to invent the “next big thing”? The answer is they don’t want to get old. Aswath Damodaran talks about this in The Corporate Life Cycle. Despite their many benefits, some companies cannot stand the notion of issuing dividends and maximizing profits, as that’s what old companies do. Young companies, on the other hand, plow billions into science experiments and R&D. They don’t care about losing money because their golden years are ahead of them. These are the types of company Big Tech wants to be, but they neglect the important and immutable fact that they’re old.
When you refuse to admit your age, bad things happen. We see this in humans: What starts with “just a little” lip-filler devolves into a slew of unaffordable botox operations that make you look even older. At a certain point, you become unrecognizable, even to yourself, leading to a series of even more questionable decisions. In the world of Big Tech, that’s what VR headsets and the “metaverse” were: a middle-aged company’s multi-billion-dollar attempt at appearing young again — the corporate equivalent of plastic surgery.
The Ultimate Face-Lift
No face-lift has been more appealing to Big Tech, however, than the one that revealed itself in late 2022: ChatGPT. For the first time in years, the world was actually excited about a new technology. And while Big Tech technically hadn’t come up with it, nothing was stopping it from bolting it onto its face and saying it did. So that’s what they did: Microsoft bought 27% of OpenAI, and Amazon and Google bought more than 20% of Anthropic. That way they might be viewed not as “Big Tech companies” but “AI companies.” Perhaps the world might see them as the sexy young startups they once were.
This is why AI failing is not an option for Big Tech. It’s not that they’d “die” — it’s just that they’d be revealed as mature. Take Microsoft’s most recent quarter, for example. Without the “AI” cloud business, the numbers look more like those of a Chevron earnings report. Commercial 365 revenue slowed to 14% growth, while Windows and Xbox revenue fell by 7% and 10% respectively. Those numbers aren’t bad — they’re just what you’d want to see from a startup. But that’s the thing: Microsoft isn’t a startup anymore; it’s a 51-year-old corporation.
The End Game
Despite the wrinkles, old age has plenty of benefits — the greatest of which is, in my view, the luxury not to care what other people think. It’s nice that I, as a young person, can outrun a middle-aged man. But it’s also somewhat annoying that my earnings trajectory largely depends on my ability to establish professional credibility and get my peers/employers to like me. In sum, what people think of me matters. But for an older person who’s already established themselves, such obligations don’t exist.
This is why I believe one of the greatest mistakes you can make is to reject your age. In the desperate pursuit of what you once had, you inevitably lose out on everything that could be yours today.
This is what I see in Big Tech. Every dollar vociferously plowed into a data center or an AI lab is a dollar that could be distributed to shareholders and spent on … anything else. It doesn’t have to be this urgent, this obsessive. Life could be simpler, easier, more calm.
It’s time for Big Tech to put down the scalpel and embrace its age. Sure, it might lose some of its multiple. But the good news is what it lost in future growth, it would make up for with dignity.
See you next week,
Ed
This may not be a message Mark Zuckerberg is ready to hear...
The capex orgy is not going to slow down until something serious happens. Once the hyperscalers started down this path, there’s no offramp for this money pit of epic proportions. FOMO is the driving factor here. The payoff has such huge potential that these companies jumped on the wave without having a plan. It was all in immediately rather than a ramp up. Whether they realized the scale of the money furnace they’ve invested in is debatable, but what is not debatable is that it is a MASSIVE money furnace. What is telling from all of the circular financing and lack of breakout of AI revenue in quarterly reports, is that the money pit is getting larger and larger with no payoff in sight.