Meta shares plummeted almost 10% in extended trading on Wednesday after the company issued a weaker-than-expected revenue forecast and took a big hit to its cash pile.
Here's how the company did, compared with estimates from analysts polled by LSEG:
- Earnings per share: $6.18 vs. $7.22
- Revenue: $60.80 billion vs. $60.17 billion
Meta said it expects revenue this quarter of between $61 billion and $64 billion, or $62.5 billion at the middle of the range. Analysts were expecting guidance of $63.15 billion, according to LSEG. The company said that the guidance "assumes foreign currency is an approximately 1% headwind to year-over-year total revenue growth, based on current exchange rates."
Daily active people, or DAP, came in at 3.6 billion, trailing Wall Street estimates of 3.61 billion, according to StreetAccount. DAP measures the number of users of Meta's family of apps.
For capital expenditures, Meta narrowed its guidance for the year to between $130 billion and $145 billion from a prior range of $125 billion to $145 billion. With Meta pouring money into artificial intelligence infrastructure, Meta's free cash flow dwindled to $784 million in the quarter from $8.55 billion a year earlier.
Meta shares are down 11% for the year as of Wednesday's close, while the Nasdaq is up about 5% over that stretch.
Investors are paying close attention to Meta's efforts to more directly monetize its various AI-related efforts. Earlier this month, Meta debuted the Muse Spark 1.1 model, which AI chief Alexandr Wang said represents the "strongest model for agentic and coding work yet" and at a cheaper price than offerings from OpenAI and Anthropic.
The company has been aggressively investing in a new AI strategy since hiring Wang in June 2025 in a deal that involved a $14.3 billion investment in Scale AI, Wang's startup.
"Overall, we expect that a significant portion of our compute is going to go towards training our models, growing our core business, and delivering personal agents and new products," Meta CEO Mark Zuckerberg said on Wednesday's earnings call. "But we also expect to grow a large business serving large customers as well."
Last week, rival Alphabet reported that free cash flow turned negative for the first time on record due to the Internet giant's hefty AI spending. Unlike Alphabet and fellow hyperscalers Amazon and Microsoft, Meta doesn't have a thriving cloud-computing business. That could be changing as Meta looks to lease out excess capacity to third parties.
"We're getting a lot of offers for compute at a significant premium over what we paid for it," Zuckerberg said on the call.
On Tuesday, Meta announced a venture with BlackRock to create a $14 billion data center project in El Paso, Texas, just a few weeks after disclosing that its big Hyperion data center project in rural Louisiana would cost over $50 billion. Earlier in July, Meta revealed plans to build a $9 billion data center in Alberta, Canada.
Meta said second-quarter total costs and expenses came in at $42.03 billion, representing an increase of 55% from a year ago. That figure includes $2.4 billion of charges related to legal proceedings and $1.18 billion of severance expenses in connection with the company's layoffs that began in May.
Finance Chief Susan Li said on the call that, excluding those charges, operating income would have increased 9% year over year.
Net income for the period slid to $15.85 billion from $18.34 billion, or $7.14 per share, in the same quarter last year.
Meta's Reality Labs unit generated $4.6 billion in second-quarter operating losses while bringing in $431 million in sales. Wall Street was expecting that unit, which develops virtual reality and related AI-powered wearable devices, to record a second-quarter loss of $5.07 billion on revenue of $423.4 million.
WATCH: AI and capex in focus for Meta this week: Here's what to expect.
Facts Only
Meta shares fell nearly 10% in extended trading on Wednesday.
Second-quarter earnings per share were $6.18; analysts estimated $7.22.
Second-quarter revenue was $60.80 billion; analysts estimated $60.17 billion.
Revenue guidance for the current quarter is between $61 billion and $64 billion.
Daily active people (DAP) totaled 3.6 billion.
Annual capital expenditure guidance is between $130 billion and $145 billion.
Quarterly free cash flow was $784 million, compared to $8.55 billion in the prior year.
Total costs and expenses for the second quarter were $42.03 billion.
Expenses included $2.4 billion in legal charges and $1.18 billion in severance costs.
Net income was $15.85 billion, down from $18.34 billion in the previous year.
Reality Labs recorded $431 million in sales and an operating loss of $4.6 billion.
Meta announced a $14 billion data center project in El Paso, Texas, and a $9 billion project in Alberta, Canada.
Executive Summary
Meta's recent financial performance presents a paradox of beating revenue estimates while triggering a sharp decline in share price. The market reaction is driven primarily by a significantly weaker-than-expected revenue forecast and a dramatic reduction in free cash flow, which plummeted from $8.55 billion to $784 million. This liquidity squeeze is the direct result of aggressive capital expenditures focused on artificial intelligence infrastructure, with annual spending projected up to $145 billion.
The company is currently attempting to pivot its AI strategy toward direct monetization. While Meta lacks the established cloud-computing revenue streams of rivals like Amazon and Microsoft, leadership suggests a new path by leasing excess compute capacity to third parties. However, these efforts coexist with substantial losses in the Reality Labs division and high one-time costs related to legal proceedings and layoffs. The central tension remains whether the massive investment in "agentic" AI and data centers will yield a sustainable return on investment before the cash pile dwindles further.
Full Take
The strongest version of this narrative is that Meta is executing a high-stakes "all-in" bet on AI infrastructure to avoid becoming a legacy platform. By absorbing massive short-term losses and slashing free cash flow, the company is attempting to build a proprietary moat in compute capacity and agentic models that can eventually be monetized both internally and via third-party leasing.
The root cause here is the "Hyperscaler Arms Race." The paradigm assumes that AI dominance is a function of raw compute power and data center scale—a capital-intensive game where the winner takes most. This echoes the infrastructure build-outs of the early internet era, where massive spending preceded a period of consolidation. The unstated assumption is that AI agents will create a new, massive revenue category capable of offsetting the multi-billion dollar burns of Reality Labs and AI training.
The human cost is reflected in the $1.18 billion severance expenses, signaling that the transition to an "AI-first" company requires the displacement of human labor to fund silicon and electricity. This represents a shift in corporate value from human capital to compute capital.
Bridge Questions: If AI monetization remains sluggish while capital expenditures continue to rise, what is the "break point" for investor patience? How does the lack of a native cloud business fundamentally change Meta's risk profile compared to Microsoft or Google?
Counterstrike Scan: A coordinated campaign to crash the stock would emphasize the "cash bleed" and the "Reality Labs sinkhole" while ignoring the revenue beat and the potential of the compute-leasing model. The current content is balanced reporting of financial data and does not match a targeted manipulation pattern.
Patterns detected: none
Sentinel — Human
The text reads like a standard, fact-heavy financial news report that synthesizes earnings data and corporate strategy, exhibiting characteristics typical of professional journalistic writing.