Meta Platforms entered the second half of 2025 with a rapidly growing advertising business that gave CEO Mark Zuckerberg significantly more financial room to pursue an increasingly expensive artificial intelligence strategy.
For the quarter ended June 30, 2025, Meta reported revenue of $47.52 billion, up 22% from a year earlier, while net income climbed 36% to $18.34 billion. The results showed how the company's established Facebook and Instagram advertising operations were continuing to generate the cash needed to fund large investments in AI infrastructure, researchers, models, and products.
Meta's Q2 2025 results were stronger than the original numbers suggested
According to Meta's official second-quarter 2025 results, costs and expenses increased 12% year over year to $27.08 billion. Operating income nevertheless rose 38% to $20.44 billion, producing a 43% operating margin.
| Q2 2025 metric | Result | Year-over-year change |
|---|---|---|
| Revenue | $47.52 billion | +22% |
| Net income | $18.34 billion | +36% |
| Costs and expenses | $27.08 billion | +12% |
| Capital expenditures | $17.01 billion | — |
Meta's Family of Apps also averaged 3.48 billion daily active people in June 2025, up 6% year over year. Advertising remained overwhelmingly dominant: the company generated $46.56 billion of advertising revenue during the quarter.
Ad impressions across Meta's Family of Apps increased 11%, while the average price per ad rose 9%. That performance mattered because Meta was simultaneously committing tens of billions of dollars to the computing capacity required for increasingly sophisticated AI systems.
Zuckerberg's goal shifted toward “personal superintelligenceâ€
On July 30, 2025, the same day Meta reported its quarterly results, Zuckerberg published a statement outlining what he called Meta's vision for personal superintelligence.
Rather than describing a finished product, Zuckerberg presented a long-term ambition: highly capable AI systems that could understand an individual's goals and context and help with work, creativity, communication, and everyday tasks. He also argued that future personal devices, particularly AI-enabled glasses, could become an important interface for those systems.
That distinction is important. Meta had not demonstrated human-level or superhuman general intelligence when the statement was published. “Superintelligence†represented the company's stated research and product direction, not an already available capability.
Zuckerberg also acknowledged that more powerful systems could introduce new safety risks and said Meta would need to be selective about what it releases openly, a notable qualification given the company's earlier emphasis on broadly available Llama models.
The AI push required much more than training new models
Meta's spending plans showed the scale of the infrastructure behind the strategy. In July 2025, the company expected full-year capital expenditures of $66 billion to $72 billion. It also warned that infrastructure costs and employee compensation would put additional upward pressure on expenses in 2026.
AI development at this level requires large fleets of accelerators, servers, networking equipment, data-center capacity, power, cooling, and highly specialized researchers and engineers. The cost therefore extends well beyond the expense of training an individual large language model.
Meta was also restructuring its AI organization after Llama 4 received a mixed reception and competition intensified among Meta, OpenAI, Google, Anthropic, and other AI developers.
One of Meta's most significant moves was a $14.3 billion investment in Scale AI in June 2025. The transaction gave Meta a 49% stake in the AI data company, while Scale CEO Alexandr Wang joined Meta to take a leading role in its new superintelligence effort.
Reports of exceptionally large recruiting offers also illustrated the competition for specialized AI talent. OpenAI CEO Sam Altman said in June 2025 that Meta had approached OpenAI employees with signing bonuses of as much as $100 million. That figure should be treated as an attributed claim rather than a confirmed standard Meta compensation package; Meta did not publicly substantiate the claim at the time.
Why Meta could afford such aggressive spending
The central advantage behind Zuckerberg's strategy was Meta's profitable advertising engine. Unlike an AI startup that must continually raise outside capital, Meta could finance much of its AI development from operations generated by Facebook, Instagram, WhatsApp, Messenger, and related services.
The company was also using machine learning and generative AI inside that existing business, including recommendation systems, advertising tools, content generation features, and automated campaign optimization. That created a potential feedback loop: improvements that strengthened engagement or advertising performance could help fund the next generation of AI infrastructure.
The Q2 2025 earnings reaction suggested investors were initially comfortable with that trade-off. Meta shares rose roughly 9% in extended trading after the report as investors focused on its strong revenue growth and third-quarter outlook, even as capital spending continued to rise.
Later results showed how expensive the strategy became
The scale of Meta's AI investment became much clearer over the following year. By the quarter ended June 30, 2026, the company's official results showed revenue had grown another 28% year over year to $60.80 billion.
However, costs and expenses increased 55% to $42.03 billion, while net income declined 14% from the unusually strong year-earlier quarter to $15.85 billion. Capital expenditures alone reached $31.08 billion during Q2 2026.
Meta also projected $130 billion to $145 billion of capital expenditures for full-year 2026, roughly double the range it had expected for 2025 when the original story emerged.
Those later figures reinforce the central tension in Meta's strategy. Its advertising business has continued to expand, giving the company enormous resources to invest in AI, but building the infrastructure and research organization required for its superintelligence ambitions has also become substantially more expensive.
The key question is no longer whether Meta can spend on AI
Meta's 2025 earnings established that the company had the financial capacity to compete aggressively in artificial intelligence. Its subsequent spending demonstrated that Zuckerberg was willing to use that capacity on a massive scale.
The longer-term question is whether those investments produce enough value to justify their cost. Some benefits can already flow through Meta's existing advertising, recommendation, and consumer products, while more ambitious ideas such as broadly available personal superintelligence remain longer-term goals whose commercial and technical outcomes are still uncertain.
For Meta, strong profits did more than support an AI experiment. They gave the company the ability to turn AI into one of the largest investment programs in its history.
0 Comments