Meta’s shares dropped after the social media giant reported second-quarter results that exposed the financial weight of its enormous AI spending.
The company reported revenue of $60.8 billion for the three months to June 30, up 28 percent year on year.
However, net income fell 14 percent to $15.85 billion, while diluted earnings per share came in at $6.18, below analysts’ $7.22 expectation.
The key issue for investors was not demand for Meta’s advertising business – but the accelerating cost of the company’s response to the AI arms race.
Meta said it now expects 2026 capital expenditure, including principal payments on finance leases, to land between $130 billion and $145 billion. That raises the bottom of the company’s previous $125 billion to $145 billion range. Quarterly capital expenditure reached $31.08 billion, helping pull free cash flow down to $784 million from $8.55 billion a year earlier.
Its third-quarter revenue forecast also landed slightly below Wall Street expectations. Meta guided to $61 billion to $64 billion in revenue, versus an analyst consensus of $63.15 billion.
Speaking to UC Today, Matt Maher, CEO of tech R&D firm M7 Innovations, said the initial direction of the earnings call made the market response unsurprising.
“The second I heard Mark Zuckerberg starting to promote his op-ed piece in the Wall Street Journal about democratizing the AI arms race and giving his company a fair shot to compete, I knew that the earnings weren’t going to be great,” Maher said.
“They beat expectations on revenue. The market just can’t handle their capex spend and their anemic cash flow, which is on life support compared to last year.”
AI Returns Remain a Future Proposition
Meta CEO Mark Zuckerberg has argued that AI is already improving the company’s core products, including content recommendations, user engagement and advertising performance.
But the results underline a familiar tension facing the largest technology companies: AI investment is immediate, while the financial return is harder to measure in a quarterly earnings report.
Meta is building an AI infrastructure footprint on a scale typically associated with cloud hyperscalers.
On the call, Zuckerberg said Meta was receiving offers for computing capacity at a “significant premium” to what the company paid for it. That is important because Meta has historically lacked the large enterprise cloud business that helps Microsoft, Amazon and Google support their own AI infrastructure investment.
“Regardless, the stark truth is that the ROI on AI investment is not ready for a Q2 earnings report,” Maher said.
“Because that can’t be shown immediately, they’re going to get punished like they are right now.
“They’re going to have to take continual beatings until they can come out on the other side and show that all these investments are actually going to pan out.”
Meta said it expects AI to serve its core business, new consumer products such as personal agents, and a future business serving larger organisations. Its proposed enterprise offering could include models, APIs, coding tools, productivity services and compute capacity.
Compute Leasing Could Open An Enterprise Route
One of the most closely watched indications of that strategy is a reported potential agreement for Anthropic to lease Meta computing capacity.
Reuters previously reported that the arrangement could be worth up to $10 billion over two years, although neither Meta nor Anthropic has announced a completed deal. If realised, it could show that Meta’s infrastructure buildout has potential value beyond supporting its own models and products.
Maher sees that possibility as an encouraging signal amid the negative reaction to the earnings report.
“If I’m being optimistic, I think that Meta actually has a ton of strong irons in the fire,” he said. “The $10 billion Anthropic deal to rent Meta’s compute is a strong indicator that Meta’s capex could be rented at a higher premium than they paid for it, which Zuckerberg actually noted.”
For enterprise technology buyers, this is the more consequential part of the company’s AI strategy.
Meta is trying to become more than a company that uses AI to power consumer feeds and sell better-targeted ads. Its roadmap points toward a role as an AI infrastructure and tools provider, taking it into more direct competition with the likes of Microsoft, Google, OpenAI and Anthropic.
Advertising Remains Meta’s Essential Safety Net
While Meta pursues new AI revenue opportunities, its advertising business remains the source of its financial power and its strategic risk.
Advertising accounts for approximately 97 percent of Meta’s revenue, Maher said. Meta’s ability to continue funding massive investment therefore rests on keeping Facebook and Instagram’s ad businesses exceptionally effective.
“Lest we forget, 97 percent of all Meta’s revenue comes from advertising,” Maher said. “They have a digital duopoly with Google. Combined, they own over 50 percent of the digital advertising market.”
AI-generated advertising and new advertising formats within Meta AI could create a more immediate commercial return than enterprise models or agents. Meta has already positioned AI as a way for advertisers to automate creative production, improve targeting and generate stronger campaign performance.
“I also think they’re making serious headway with their AI-generated ads and ad units within Meta AI, which is unproven territory that could show a growth story within their advertising ecosystem,” Maher said.
But the concentration of Meta’s revenue also explains why diversification is central to its strategy. Google remains heavily dependent on advertising, but it has more substantial cloud, subscription, hardware and enterprise businesses than Meta.
“The only difference is Google generates 75 percent of its revenue from advertising, so it’s more diversified than Meta at 97 percent,” Maher said. “They need that to continue to be so unbelievably strong they actually diversify.”
