Meta and Microsoft Q2 2026 earnings: AI capex surge and investor reaction

Meta's Q2 2026 revenue grew 28% year-over-year to $60.8 billion, comfortably beating expectations ^[1]. But that growth was overshadowed by a near-total collapse in free cash flow, which fell 91% year-over-year to just $784 million, down from $8.55 billion in the same quarter a year earlier ^[2]. Investors reacted immediately, sending Meta shares down roughly 10% in after-hours trading ^[2].

The mechanism behind the collapse is straightforward. Meta's Q2 capital expenditure jumped to $31.1 billion, more than double the $17.0 billion it spent in the same period a year earlier ^[1]. Facing a widening bill for AI infrastructure, Meta narrowed and effectively raised its full-year 2026 capex guidance to a range of $130-145 billion ^[4]. CFO Susan Li attributed the increase to higher component pricing and additional data center costs ^[5], spending built ahead of future-year capacity needs ^[13], an admission that spending is running ahead of realized revenue rather than tracking it.

CEO Mark Zuckerberg pushed back directly on skepticism about the payoff, noting that Meta is fielding outside offers to buy its compute at a significant premium to what it paid, but is choosing not to sell ^[3]. He called it foolish to sell off compute for a short-term profit instead of reserving it for Meta's own AI roadmap, adding that he personally believes AI investors are going to be rewarded and feel very good over time ^[3].

Microsoft reported its fiscal Q4 2026 results the same night as Meta, and the market's reaction could not have been more different. Azure cloud growth exceeded the 39.6% analysts had expected, helping send Microsoft shares up roughly 2% after the report ^[6]. Net income for the quarter ended June 30, 2026 rose to $35.77 billion, or $4.81 per share, up from $27.23 billion a year earlier ^[7].

Microsoft is spending just as aggressively as Meta, if not more so. It guided calendar-year 2026 capex to roughly $190 billion and disclosed FY2027 capex guidance of $255-260 billion, a roughly 35% jump ^[8]. CFO Amy Hood attributed about $25 billion of the increase to higher component pricing ^[5], echoing Meta's own explanation for its ballooning capex. The difference is that Microsoft's spending maps visibly onto Azure's accelerating growth rate, giving investors a legible return story that Meta's internal-model-training narrative does not yet offer.

Analyst Thomas Monteiro summarized the pattern investors are now applying across the sector, noting that Meta's report echoes what markets saw from Alphabet and Tesla the prior week: strong revenue growth, but even faster growth in spending ^[2]. That gap between spending growth and revenue growth, not the absolute size of the capex number, appears to be the variable driving whether a stock rises or falls on earnings night.

Meta and Microsoft's earnings landed inside a much larger pattern. Combined 2026 capital expenditure across Alphabet, Microsoft, Meta and Amazon is projected to reach up to $725 billion, a 77% increase from the prior year ^[5][9], a figure commentators noted exceeds the GDP of a country like Switzerland or Turkey.

Alphabet had already previewed the market's new skepticism less than a week earlier: on July 23, 2026, it raised its 2026 capex guidance to $180-190 billion citing unprecedented AI compute demand, and its stock still fell roughly 7% despite record profit ^[10]. Amazon, by contrast, held its 2026 capex guidance steady at $200 billion ^[5]. Meta itself had already signaled this trajectory back in April 2026, when it first raised full-year capex guidance to $125-145 billion and saw shares drop 6-7% despite a revenue beat ^[11], meaning the market's punishing reaction to Meta's Q2 print was the continuation of a months-long repricing of AI capex risk that CNBC had already flagged as an industry-wide cash-flow squeeze back in February ^[12].

Beyond the immediate stock reactions, the scale of hyperscaler spending is starting to raise structural concerns. Moody's vice president Kevin McNeil warned that even large, established players are embarking upon almost unprecedented capital investment that could pressure credit quality as they shift from asset-light to asset-heavy balance sheets ^[10].

UBS analysts separately flagged limited visibility into hyperscaler capex plans beyond 2027 as a continuing overhang on investor risk appetite, even where near-term AI demand looks solid ^[10]. That uncertainty is compounded by the fact that a meaningful share of the new spending is not purely demand-driven: both Meta and Microsoft cited rising component and memory pricing, not just AI compute demand, as a driver of their higher guidance ^[5], while Meta separately pointed to additional data center costs to support future-year capacity, meaning some of the spending is being built ahead of, rather than in direct response to, realized demand ^[13].

Retail investor communities have converged on a no-win framing for hyperscaler capex: raising it fuels fear of margin destruction, while cutting it would signal the AI buildout itself isn't working. Trading forums focused on memory-chip suppliers, which stand to benefit most directly from hyperscaler capex, read Meta and Microsoft's results as validating continued chip demand even as sentiment toward Meta's own stock stayed cautious, with one active thread debating whether Microsoft can throttle spending more freely than Meta given the different shape of their AI product roadmaps.

On social platforms, chatter around Meta's results skewed skeptical, drawing unfavorable comparisons to Microsoft's more explicit commitment to staying free-cash-flow positive despite its own rising capex, a contrast some read as an implicit dig at Meta's cash-flow trajectory. Commentary among finance-focused video channels framed the market as newly bifurcating hyperscalers by capex intensity, rewarding capex-lite business models and punishing heavy spenders unless the spending shows up in growth metrics like Azure's, which is roughly how the Meta-Microsoft divergence played out in practice.