Meta would have had a great quarter. That is, if it weren’t for the layoffs, the lawsuits and the galactic-level AI operating costs.
The company’s Q2 earnings call on Tuesday evening started off strong, as CFO Susan Li announced quarterly revenue totaling $60.4 billion, up 28% from last year. Ad revenue across Meta’s family of apps was up 27% year over year and accounted for $59.4 billion of the quarterly haul. That last $1 billion came from a delightfully specific “other revenue” column.
However, the starkest year-over-year increase came in the form of Meta’s expenses, which were up 55% for a total of $42 billion.
Li didn’t shy away from explaining why: Meta faced a combination of severance expenses ($1.2 billion) after laying off 8,000 employees in May, as well as ongoing legal proceedings ($2.4 billion).
Using the exact same verbiage as she did in Q4, Li briefly mentioned the “number of youth-related trials” that “may ultimately result in a material loss” for Meta – although one of its most high-profile cases, in which a teenager alleged that Meta created an addictive and harmful environment, was dropped last week.
Li said the other biggest factor in the expense increase was rising AI costs, including third-party tokens and data center development, as well as employee compensation growth. Most of the new hires were technical hires, she said, “particularly AI talent.”
No shocker there.
Right on target (almost)
Meta’s hiring of high-price AI talent may also be the reason that Meta’s advertising business saw so much growth in recent months. CEO Mark Zuckerberg told investors that Meta’s LLMs are able to understand the subject matter of a piece of content and “why it is compelling,” which provides clearer insights into people’s interests and goals based on the content they engage with. This allows the system to determine an ad’s relevance and to predict or rank the ads that a user should see.
He also highlighted the recent launch of Muse Image, a generative AI tool designed to analyze and edit images and create ad variations. He neglected to mention the short-lived feature that allowed users to generate AI-likenesses of any adult users with public accounts who hadn’t specifically opted out. The feature lasted three days before Meta retracted it, citing that it had “missed the mark.”
According to Zuckerberg on Wednesday, Meta has been “getting great feedback on [Muse] so far.”
Looking forward, Meta plans to develop foundation AI models to power organic content and ad recommendations at the same time, said Li, as well as “LLM-native recommender systems.”
And Meta’s ads better be top tier, because users are going to start seeing them everywhere.
In Q2, the company completed its global ads expansion on Threads, for what that’s worth, and continued its rollout in WhatsApp.
In early tests, Meta’s new AI tools (including deploying a generative model in its ads retrieval system) led to an 8.3% increase in ad clicks and a 15.7% uplift in conversions on Facebook.
No such thing as free AI
Meta is betting a great deal that its AI tools will pay off not just in the coming months but for years down the line.
Zuckerberg repeatedly referenced the importance of building new data centers – indeed, on Tuesday, Meta announced a venture with BlackRock to develop a new one-gigawatt center in El Paso, Texas – to match “the pace of AI adoption.”
Some investors were skeptical.
In addition to the lofty costs, there’s growing concern about the social and political backlash against data center development. Recently, Gov. Kathy Hochul announced a moratorium on new data centers in New York State.
One investor asked how Meta was thinking about the costs “philosophically.” Which, when the question comes from an investor, can be confidently assumed not to refer to concerns over potential environmental impacts.
“How are you guys thinking about wanting to be ambitious on the spend, but then marrying that with the need for capital?” he asked.
Apparently, Meta has also shifted its capital structure in recent years to reduce the burden on investors in its long-term, pricey initiatives.
The new structure, said Li, is “a greater mix of debt.”
