Mark Zuckerberg is trying to buy the future

Right now, the Meta $META CEO — and his company — is spending like a man who has never heard the word “budget.” The Silicon Valley giant, in its second-quarter earnings posted Wednesday, hiked its 2025 capital expenditure forecast to an eye-watering $66-72 billion, close to triple its 2022 spending and more than double what it initially planned for this year. That money is being poured into “superintelligence” labs, custom silicon, AI data centers, high-performance compute clusters, and the kind of recruiting benders that are usually reserved for startup fever dreams.

Zuckerberg has reportedly offered nine-figure compensation packages to lure top researchers from OpenAI, Google $GOOGL DeepMind, and other competitors, fueling an AI arms race so intense that Meta’s AI budget now rivals that of sovereign nations.

Wall Street’s response? Applause. Keep going. You’re doing great, sweetie! Wedbush analyst (and major tech bull) Dan Ives wrote in a post on X $TWTR: “We have been talking about the AI Revolution for years but tonight was a night the tech world and the Street will remember for a long time.”

On paper, this is one of the most aggressive infrastructure buildouts in Silicon Valley history. In practice? It’s the cost of keeping up with a narrative that Zuckerberg really, really doesn’t want to lose: That Meta is still a technology company, not just an ad platform with delusions of grandeur. But that core business — digital advertising — remains such a relentless moneymaker that, as long as the clicks keep converting, Zuckerberg gets to keep playing visionary. And investors get to keep cashing in.

The company’s second-quarter earnings report was a financial flex: $47.52 billion in revenue (up 22%), $7.14 in EPS (up 38%), and operating margins climbing to 43%, even as Meta dropped another $3.9 billion into Reality Labs, a business arm that only Zuckerberg seems to actually care about (and one that investors have all but given up trying to model a payback from, but tolerate because ad margins are so rich).

Meta has quietly built a solid AI stack — from custom silicon to open-source LLMs — and insists these investments will eventually unlock entirely new revenue streams, from assistant bots to smart glasses to enterprise APIs. The timeline? Flexible. The price tag? Rising. But if this is a bet on Meta’s future as an AI-first company, it’s being bankrolled by the past. 

The cash cow behind the curtain

Meta still makes 98% of its revenue from ads (per second-quarter numbers), and the company’s model has only gotten more efficient. Thanks to AI-fueled targeting, Meta’s ad tools are delivering more impressions at higher prices, all with better conversion rates. It's working: Meta’s ad engine is now doing more with less. Thanks to a suite of AI upgrades — GEM, Andromeda, Lattice — ad conversions improved by 5% on Instagram and 3% on Facebook, while video watch time grew 20% across both platforms. Nearly 2 million advertisers are reportedly now using Meta’s AI tools to auto-generate text and images, and “click-to-message” revenue grew 40% year-over-year.

Jeffries analysts wrote in a Thursday note that AI “enabled greater efficiency gains across Meta's ad ecosystem,” calling out the AI-powered recommendation model for ads. That’s Wall Street speak for: The dog food ads are getting harder to skip, and Meta’s getting paid more when you click.

“Facebook is the digital advertising juggernaut,” said Motley Fool senior investment analyst David Meier. “You have three and a half billion people touching one of your apps every day. You’re able to find ways now using AI tools to help you do it. You’re finding ways to get 11% more ad impressions in front of all those people. You have such an effective ad algorithm that generates conversions that you can structure higher prices than last year.”

Advertising isn’t sexy, but it pays the bills — and then some. 

That creates a strange tension: Zuckerberg seems to want to frame Meta as a cutting-edge AI company… while still depending almost entirely on the same business the company ran on a decade ago. Meta’s flashy AI models could one day shape the future, but right now, it’s the company’s tried-and-true algorithm quietly placing ads between Instagram Reels that is funding Meta’s attempted reinvention.

“They are the monster,” Meier said. “They have the platform. They have the company they want. They have people willing to see ads, and advertisers get those ads to convert, I mean, they just keep printing money.”

Wall Street likes results, and Meta gave them plenty. Revenue beat expectations by 6%, operating income by $3.4 billion. Wedbush called the upside “magnitude,” while Deutsche Bank praised Meta for flagging “durable top-line growth.” All four banks raised their price targets to between $920 and $950. William Blair analysts noted the “massive” quarter in an analyst note and wrote that “the company is demonstrating that it can profitably scale the business while making massive capital investments. … We continue to be positive on its AI adoption and benefits to advertisers and believe Meta will be a long-term AI leader.”

Analysts from across the board echoed a similar sentiment: Meta’s ad dominance gives it runway to spend — and win. 

Jefferies sees the capital expenditures surge as a sort of strategic moat, praising Meta for capital investments that are “well aligned with [long-term] growth initiatives.” Wedbush said that “the infusion of AI capabilities across the company’s ad stack and content recommendation engines are driving tangible results.” Deutsche Bank noted Meta is “leaning in to foundational AI model training work, which is unlocking new vertically integrated opportunities.” And Cantor Fitzgerald echoed that framing, adding that the company is “well-positioned with capital, infrastructure, and talent to achieve the next technological breakthroughs.”

But the checks aren’t blank forever, and there could be growing concerns about just how sustainable Meta’s investment pace really is. The company isn’t just guiding toward $72 billion this year; it’s already forecasting another jump in capital expenditures next year, with chief financial officer Susan Li suggesting spending could ramp up to $100 billion. Cantor warned that the “stakes on AI-game theory” between megacaps could lead to runaway spending.

And before too long, Meta might have to prove the other side of its AI equation: not just building infrastructure, but monetizing it. But Wall Street doesn’t seem too stressed.

“While the scale of these ambitions is vast, and requires significant investments that have a longer-dated payback horizon, we contend that these capital investments are also having strong returns in terms of engagement and ad performance, which is translating to Meta's current durable advertising revenue growth, while simultaneously affording Meta the ability to be a leading player in both hardware and AI advances,” Deutsche Bank analysts wrote. “Given this mix of near-term tailwinds, alongside longer-term strategic advances (with ample greenfield opportunities), we contend that Meta deserves a premium valuation multiple vs peers.”

For now, Meta’s advertising business gives the company carte blanche while it claims the future will be kind to — and profitable for — the company. Still, that’s a lot of pressure on one cash cow.

$10 trillion — or bust?

Could it all work? Could Meta actually parlay its AI investments into a second act so successful that it doubles — or even quintuples — its market cap? 

“It is shocking that I’m getting ready to say I could easily see Meta doubling in five years,” Meier said. “Growing 15% a year for the next five years, based on how the company is performing now, is not a stretch.” At the moment, Meta sits close to a $2 trillion market cap, meaning Meier’s back-of-the-envelope math lands somewhere near $4 trillion by the end of the decade. If the AI bets pay off? That number could be bigger — and faster.

Of course, forecasts are easy when the revenue keeps flowing. The test for Meta is whether it can generate entirely new demand, not just squeeze more margin from an already mature business. Regulators are still watching the company closely. The Department of Justice’s long-running antitrust investigation could threaten Meta’s ability to consolidate data across its properties or, in a worst-case scenario, force divestitures of Instagram or WhatsApp. 

Meta also faces internal friction. Employee churn in the company’s AI teams has been high, and its much-hyped Llama Behemoth model has been delayed (and reportedly isn’t all that good) — the kind of technical hiccup that analysts notice. Then there’s the hardware wildcard: Meta’s bet on smart glasses as the next big computing platform. Zuckerberg says the form factor is the future, but history hasn’t exactly been kind to wearable hype.

“The other thing that’s going to be incredible to see in this next space where Meta is making these investments — in people, infrastructure, technology, development — is: How is all that going to come together?” Meier said. “Because, again, think about the price tag. $200 billion is going to flow out of the company, and [Zuckerberg] wants to generate a return on that. We as investors get a front-row seat to, OK, how is this all going to play out? 

“If it works, they’re going to be writing Harvard Business School cases about all of this.”

That may sound optimistic, but it’s not necessarily unfounded. Meta’s models are improving. Its compute fleet is growing. Its monetization tools are working. And if AI does usher in a platform shift — if search, productivity, commerce, and devices all reconfigure around bots and assistants — Meta could be well-positioned to profit. Still, this is a bet. And an expensive one. As Deutsche Bank said, Meta is “affording [itself] the ability” to be a leading player in both hardware and AI. That’s a subtle but crucial distinction.

Meta hasn’t won yet. Right now, it’s still paying to stay in the race. Whether Meta is building the future or just buying time, the one thing the company is definitely doing is spending — a lot.