So Meta reported earnings yesterday afternoon (April 24th, after the bell) and I've had one tab open on my second monitor since about 4:05pm just watching the after-hours ticker do its thing. Revenue came in at $36.46 billion, up 27% year over year, comfortably ahead of what analysts wanted. EPS beat too. By every normal measure that's a good quarter.
The stock still dropped. Hard. Down around 15% in after-hours trading, which for a company Meta's size is not a small number, that's tens of billions of dollars of market cap gone in about twenty minutes.
The reason isn't the quarter that already happened, its the one that hasn't. Zuckerberg got on the call and basically told investors to buckle up for a lot more spending on AI infrastructure than anyone had priced in. The company raised its full-year capex guidance to a range of $35-40 billion, up from the $30-37 billion they'd said back in January, and the tone on the call made it pretty clear this isn't a one-year thing. He used the phrase "multi-year investment cycle," which is the kind of phrase that sounds fine in a press release and sends a shiver down the spine of anyone who remembers how "metaverse" investments went over the last two years.
I get why Wall Street flinched. Meta spent something like $46 billion on Reality Labs since 2020 and mostly has a mediocre headset and some so-so glasses to show for it, commercially speaking. So when the same guy gets on a call and says "trust me, this next multi-year bet is the real one," a certain amount of skepticism is earned, not just knee-jerk.
But I actually think the market's got this wrong, or at least is reacting to the wrong thing. Buying GPUs and building data centers to train and serve Llama models is a completely different kind of spend than building a VR chat app nobody asked for. One of those things has an obvious, immediate, revenue-generating use case (better ad targeting and ranking models, which is literally the entire Meta business), and the other was a bet on hardware adoption curves that take a decade to play out, if they play out at all. Lumping "AI capex" in with "metaverse capex" because they're both big numbers from the same CEO feels lazy to me, honestly. I'm not saying give the guy a pass because I like Llama 3, which came out last week and which I've actually been running the 8B version of locally on my machine and it's genuinely good for its size. I'm saying the two spending sprees aren't the same animal even if the headline number rhymes.
Threads also crossed 150 million monthly actives on the call, which got mentioned almost as an aside, which is a little funny given how much oxygen that launch took up a year ago. Feels like it's settled into "fine, exists, not going anywhere, not really threatening Twitter/X either" territory, and I think that's basically where most of us figured it would land.
Anyway. None of this is investment advice, I don't own Meta stock and don't plan to, my one brush with meme-stock energy was a very small and very regretted GameStop position in 2021 that I will not be discussing further. But I do think this is worth watching over the next two or three quarters, because the actual test isn't whether Zuckerberg can spend the money, it's whether that spend shows up anywhere except the AI research papers. If Meta's ad business gets measurably better at prediction because of all this new compute, the market will forgive the capex number fast. If it's another few years of "trust the roadmap," people are going to remember how the Reality Labs bet went and stop being patient.
One small complaint while I'm here: can companies please stop calling every large expenditure an "investment cycle"? It's spending. Call it spending. Investment cycle is what you say when you want the number to sound temporary and strategic instead of just large.