Meta AI spending has become the defining story of this earnings season. The company posted record revenue for the second quarter and still watched its profit shrink, a combination that has forced Wall Street to confront the real cost of Mark Zuckerberg's superintelligence race.

A Record Quarter Investors Hated

Meta AI spending turned what should have been a victory lap into a warning. On July 29 the company reported second quarter revenue of $60.8 billion, a 28% jump from a year earlier and comfortably ahead of Wall Street forecasts. Advertising, still the engine of the entire business, delivered $59.4 billion of that total, with ad impressions up 14% and the average price per ad up 12%. By almost any historical standard, this was a strong three months.

The user base held up well. Family daily active people averaged 3.60 billion in June, up 3% year over year, proof that engagement across Facebook, Instagram, WhatsApp and Messenger remains remarkably sticky at enormous scale. Demand for Meta's advertising inventory is not the problem here.

And yet the stock fell in after-hours trading. The reason sits one line further down the income statement. Net income dropped to $15.8 billion from $18.3 billion a year earlier, a 14% decline, and diluted earnings per share slipped to $6.18 from $7.14. For a company investors had grown used to seeing compound profits quarter after quarter, a falling bottom line during a revenue boom was jarring. The culprit was not weak demand. It was cost.

Where the Money Went

Total costs and expenses reached $42.0 billion, a 55% increase year over year that dwarfed the 28% rise in revenue. When your bills grow at twice the pace of your sales, margins compress fast. Meta's operating margin fell to 31% from 43% in the same quarter last year, and operating income actually declined 8% despite the record top line. Meta AI spending explains almost all of that gap.

Research and development was the standout. That line ballooned to $21.7 billion from $12.9 billion, an increase of roughly $8.7 billion driven almost entirely by the company's artificial intelligence ambitions and an aggressive hiring push for its superintelligence teams. Capital expenditure, including finance leases, hit $31.1 billion in the quarter alone. Meta AI spending is no longer a rounding error inside a hugely profitable ad business. It has become the defining feature of the company's financial profile.

There were other one-time weights too. Meta booked $2.4 billion in charges tied to legal proceedings and $1.18 billion in severance from the May 2026 headcount reduction that affected roughly 8,000 employees. But strip those out and the core story is unchanged. The bulk of the expense growth is deliberate, strategic, and aimed squarely at building AI infrastructure.

The Cash Flow Cliff

The most alarming number never made the earnings headline. Free cash flow, the money left after running the business and paying for property and equipment, collapsed to just $784 million. A year earlier it was $8.5 billion. That is a drop of more than 90% in twelve months, and it happened even as operating cash flow rose to $31.9 billion. Everything the business generated, and then some, is being poured straight back into data centers and chips.

To keep the machine fed, Meta raised $24.9 billion in new long-term debt during the quarter, according to its financial statements filed with regulators, pushing total long-term debt to $83.7 billion. This is a company that spent years handing cash back to shareholders through buybacks. In the second quarter it repurchased no stock at all. The shift from cash generator to cash consumer is the clearest sign yet that Meta AI spending is reshaping the company's entire capital structure, not just a single quarter's results.

The balance sheet tells the same story from another angle. Property and equipment, net of depreciation, swelled to $225.7 billion from $176.4 billion at the end of 2025, a roughly $49 billion increase in half a year. That is the physical footprint of the AI buildout, the servers, land and power infrastructure that Zuckerberg believes will define the next era of computing. It is also a fixed cost that will depreciate for years whether or not the payoff arrives on schedule. Cash and cash equivalents, meanwhile, fell to $15.5 billion from $35.9 billion at the start of the year, a reminder that even a business this profitable has limits on how much it can self-fund.

Microsoft's Relief, Meta's Reckoning

The contrast with Microsoft, which reported the same evening, was stark. Microsoft posted revenue of about $90 billion, up 18%, with Azure growth strong enough to reassure investors that its own enormous capital budget is translating into cloud sales. Microsoft can point to paying enterprise customers renting its AI capacity. Meta cannot yet do the same, because Meta AI spending mostly serves its own products rather than a third-party cloud business.

That difference matters. Both companies are spending fortunes, but Microsoft has a meter running on its infrastructure while Meta is making a longer, less visible bet. Investors rewarded the company with the clearer payback and punished the one asking for more patience. For a fuller breakdown of how the hyperscaler capex race is splitting winners from worriers, see Blumefield's earlier coverage of the 2026 earnings season.

The Superintelligence Wager

Mark Zuckerberg framed the quarter as proof the strategy is working. "AI is accelerating our core business today, powering our next generation of products, and opening the door to entirely new enterprise opportunities," he said. The company raised the low end of its 2026 capital expenditure guidance to a range of $130 billion to $145 billion and now expects full year expenses of $165 billion to $169 billion.

The bull case is straightforward. Meta's AI systems already sharpen ad targeting and content recommendations, which is why impressions and prices keep climbing. If those same systems can be pushed toward what Zuckerberg calls personal superintelligence, the eventual prize could dwarf today's costs. The bear case is just as simple. Meta AI spending is real and immediate, while the returns beyond advertising remain a promise. Reality Labs, the company's other long-horizon bet, lost $4.6 billion in the quarter and offers a cautionary reminder of how long expensive visions can take to pay off.

For now the market has cast its early vote. Record revenue was not enough to offset a shrinking profit and a vanishing cash cushion. The coming quarters will reveal whether Meta AI spending is the foundation of the company's next decade or the moment ambition outran discipline. Investors, and the rest of Big Tech watching closely, will be reading every line of the income statement. Follow the story as it develops at Blumefield.