Join institutional investors, analysts and strategists from the world's largest banks: Subscribe today
Already have an account? log in
Meta Spooks Investors With Spending Plans, Cost Outlook
The bar for Meta was quite high headed into the company's Q1 results on Wednesday afternoon in the US.
The stock's up sharply so far in 2024 thanks in no small part to a singular post-earnings rally in February, when the company rode a beat and raise paired with a first-ever dividend to a one-day value creation event for the history books.
Fast forward three months and Meta said Q1 revenue was $36.46 billion, narrowly ahead of the $36.12 billion consensus. Ad revenue was $35.64 billion, a touc
From stock action during call, I think investors got spooked by the CEO’s cautioning of a multi-year investment cycle in “Meta AI” while products scale but revenues are slower to arrive and how stock has historically been “volatile” in similar periods. Others may have been off-put by no geographic DAU MAU ARPU data in the slide deck. In 1Q-3Q META moves into a period of much tougher YOY comps in some geographies, so will 1Q24’s +27% YOY revenue growth fall to +LSD or even nil by year-end? How can you tell without geo ARPU numbers?
The thing is 1Q24 YOY revenue growth actually accelerated into significantly tougher comps. Hmm. And if you reconstruct the missing geo ARPU data (it is possible, at least approximately) it looks like in every geo, 1Q24 YOY ARPU growth significantly outperformed the tougher comps. Hmm.
My speculation is that, while public-facing “Meta AI” revenue may be immaterial for a long time (kind of like Metaverse), META’s internal “Advertising AI” (made-up name, that’s not what they call it) is driving impressive under-the-covers growth in advertising targeting, delivery, and creation.
In other words, if you’re looking for a non-semi company that is actually making big revenue from AI, META may be it in spades – its just in the advertising business which is how META makes its money anyway, rather than consumers directly paying to use LlaMa 3 on SmartGlasses 2 or whatever.
I’ve never understood how CEOs manage to trip up their own shares on earnings calls. This isn’t hard: Just hop on there, say what you know everyone wants to hear, obfuscate if you get a question you don’t like and be done with it.
I added two sentences to the end of this article after listening to the snippet from the call you referenced. He shouldn’t have said that. Everyone knows it anyway. There was no reason for him to reiterate it. He was just asking for another 5% downside on the stock. And he got it.
Zuckerberg is an interesting guy. I’ve been watching, in phases, a 90 minute interview with him. He’s focused on the long view and seems much less willing to overpromise and underdeliver than, say, Musk. I’d guess he’s also licking his lips at the prospect of a TikTok ban – though I wonder if that will pass legal review.
Now that Zuck says he’s spending up to $40 billion on AI, how likely is it that AWS, MSFT, Oracle, Alphabet, IBM, trim their AI Capex? Are they likely to settle for second tier servers/GPUs/hardware that have say 70-80% of the performance of the latest and best hardware just to save a few bucks ??. How likely is it that Google search is disrupted ?? Is NVIDIA excessively valued at just below 30x forward earnings estimates?? Is Micron/MU excessively valued considering it is the only American high bandwidth memory chip producer?? I bet most if not all the cloud service providers increase their AI Capex. All roads for now lead to NVIDIA ??
An old man’s read of the recent price action in AI stocks is that the theme is losing some momentum. Secondary plays, such as PLTR, have been trading heavy for a few weeks and even the mighty NVDA and AMD have moved from relentless to volatile.
Are we moving from unicorns & rainbows (promises) to “Show me the Beef” (cost savings) when it comes to AI?
I think we are, and it’s overdue.
Show me the cost saves, and even better show me the revenue acceleration.
I wonder who will show us a revenue boost from AI (again, other than semi names).
Going back to META, 1Q rev was +27.3% YOY, accelerating from +24.7% 4Q. Sequential -1.2% was much better than past 1Q QOQs (typically -MSD/HSD). How much of the higher growth was more effective content recommendation and ad targeting, delivery, and generation from META’s internal AI (which would be machine learning as well as LLM)? It could have been a $1BN revenue boost. Speculation, of course.
MSFT needs to show some material $’s for CoPilot AI revenue. I think so, anyway.
JL – We will see real world examples but not at the pace that today’s theme-driven investors would like. Online retailing did not immediately lead to widespread brick & mortar store closures. It took a while.
Nor will all companies uniformly benefit from AI usage. If the firm is in a highly competitive industry, won’t cost savings need to be passed through to consumers to match what their competitors are doing? Not the case for the companies with near-monopolies which will be better to maintain pricing and pass the cost-reductions on to shareholders via buybacks.
Just saw a piece from Business Wire (“a Berkshire Hathaway company”).
Included in the results from a survey of CIOs was little snippet:
“What has remained consistent with previous years is IT’s ongoing challenge to measure impact. Sixty-one percent of CIOs said they find it very or extremely challenging to demonstrate return on investment (ROI) with tech investments. While 96% of CIOs anticipate increased investment over the next 12 months, 42% of respondents admit they do not expect to see positive ROI from AI investments for at least two to three years.”
Peanut brittle. Everybody should just make Peanut brittle. It requires no technology, it’s delicious and the ROI can be measured in smiles.
My first academic publication (in 1975) was concerned with the inherent difficulties in trying to measure the (cash) benefits of tech investments and how to assess their actual performance. My first consulting job in grad school was to prepare a comprehensive sales force training manual for NCR. They wanted their sales people to know about how to teach customers the ins and outs of the PV and IRR measures they could use to decide how best to acquire NCR’s tech systems. Still, few know how to cut this Gordian Knot.
JL – I hope we get some replies to your sharp questions from decision makers in the advertising space. My own VERY LIMITED experience is that GoogleAds are way more effective than FB ads. Though I do appreciate the swimsuit ads fed alongside the endless offers from Temu!!!
It is interesting to see Chinese retailers/retail apps (Temu, Shein, etc) seeking growth in the US/West while Chinese consumer spending lags. I wonder if Pinduoduo will follow.
Until they also are banned for “national security” reasons.
Maximizing clicks on facebook, etc. and continuing on the reprehensible path Frances Haugen’s described to Congress = “artificial” intelligence.
H is an example of a “smart” Meta investor (see other articles).
Zuckerberg is in the process of demonstrating how ego can lead to blowing billions of dollars. It will get worse…Meta isn’t exactly prepared for a pullback in advertising expenditures.