The AI infrastructure race is entering a phase of commercial validation. Both Meta and Microsoft are increasing investments by hundreds of billions of dollars, but after earnings reports, the market responded differently: Microsoft Azure’s growth demonstrates that AI is generating revenue, while Meta still needs to answer when its massive investments will deliver returns.
After releasing their quarterly earnings following the close of U.S. equities trading on Wednesday, the two tech giants with the most aggressive investments in AI infrastructure received entirely different evaluations from Wall Street.
After Meta (META.O) released its earnings report, investors continued to ask when the company’s hundreds of billions of dollars in AI investments would generate tangible returns; meanwhile, Microsoft (MSFT.O) demonstrated that AI infrastructure investments are delivering commercial value, driven by accelerated growth in its Azure cloud business.
Both companies face a similar core issue: whether massive capital expenditures following the AI race's entry into a large-scale investment phase will generate sufficient revenue growth. However, the market's assessment of the two is now clearly diverging.
Meta is trying to convince the market with its AI vision, but investors are still waiting for returns.
Before the earnings report, Meta CEO Mark Zuckerberg actively signaled the company’s AI strategy, aiming to demonstrate to the market the long-term value of its continued investments in AI.
On Tuesday, Zuckerberg published an op-ed in The Wall Street Journal expressing optimism about the widespread adoption of “superintelligence.” He suggested that if everyone had access to superintelligence tools, society could become more equitable and efficient than it is today.
As a thought experiment, imagine that only one person possesses a superintelligent lawyer. They would have an unfair advantage in court, leading to a worse society. But now imagine that everyone has a superintelligent lawyer—justice would be far fairer and more efficient than it is today.
Zuckerberg previously emphasized in interviews with The New York Times and the Financial Times that AI should move toward an era of "personalized superintelligence," where everyone has an AI assistant customized to their own needs and beliefs.
He believes that AI should not be centrally controlled by a few companies, but should instead develop in a more open manner. Zuckerberg said, "I think a single benevolent superintelligence that serves everyone's interests simultaneously is not possible."
However, these optimistic statements about AI's future have not alleviated Wall Street's concerns about Meta's short-term business returns.
After Meta released its second-quarter earnings report, the company's stock briefly dropped nearly 8%. According to Bloomberg, Meta's earnings per share were $6.18, below the market expectation of $7.14; revenue reached $60.8 billion, exceeding analysts' forecast of $60.23 billion.
The financial pressure stems primarily from profitability and expectations for future investments. Meta’s CFO, Susan Li, stated that the company incurred $2.4 billion in expenses during the second quarter due to legal proceedings, raising the full-year total expenditure forecast to $165 billion to $169 billion, up from the previous range of $162 billion to $169 billion.
Meanwhile, Meta raised its lower bound for 2026 capital expenditure guidance from $125 billion to $130 billion, while keeping the upper bound unchanged at $145 billion. The company has previously stated that a significant portion of this capital expenditure will be allocated to AI infrastructure.
Investors are focused on when Meta’s increasing investments in data centers, computing resources, and chip purchases will translate into new drivers of revenue growth. Over the past year, Meta’s stock has declined by approximately 10%, partly because the market is waiting for the company to demonstrate that its AI investments will generate returns.
During the earnings call, analysts directly asked Zuckerberg about the commercialization of AI.
Bernstein analyst Mark Shmulik asked when consumers will truly change their behavior due to AI. He referred to the current form of AI products as a “glorified search tool” and asked:
Are we on the brink of a breakthrough, or do we simply need more patience?
Goldman Sachs analyst Eric Sheridan is also focused on the timeline for AI return on investment. He asked Meta which business lines, given the current product lineup and computing capabilities, are best positioned to achieve scale by 2026 and 2027 and demonstrate “quantifiable, substantial return on invested capital” to investors.
Meta attempts to commercialize its AI infrastructure, but the market still has questions.
In response to external skepticism about AI investments, Zuckerberg emphasized Meta’s business model flexibility, stating that the company can leverage its computing power to develop new capabilities beyond advertising, including enterprise services and internal operational tools.
In July, Zuckerberg announced that Meta might rent out some of its idle computing capacity. The plan briefly boosted market confidence, as investors believed Meta could monetize part of its AI investments through its cloud computing business.
Reports indicate that Meta is exploring a cloud service-like business model to increase infrastructure utilization by selling AI computing power.
However, analysts still have questions about this strategy. During the earnings call, JPMorgan analyst Douglas Anmuth noted that if Meta sells computing resources to external customers while simultaneously needing to purchase computing power from third parties, the company could potentially become both a supplier and a buyer of computing resources in the future.
He asked Zuckerberg: "Mark, regarding your various approaches to commercializing external computing power, you're also purchasing computing power from multiple third parties. I'd like you to help us understand the difference."
In addition to business model issues, Meta faces challenges with user trust and regulatory pressure.
Mike Proulx, Vice President and Research Director at Forrester Research, said that Meta is currently facing "trust costs" across multiple growth initiatives.
He noted that AI-generated ads may pose brand control issues, smart glasses involve privacy risks, adolescent safety remains a regulatory concern, and employee data tracking measures have sparked controversy over data collection.
Meta currently faces approximately 3,000 lawsuits related to children's rights, filed by families, school districts, and state attorneys general. Regulatory scrutiny over the safety of social media platforms may also impact Meta’s advertising-dependent growth model.
Microsoft's AI investments are beginning to pay off, with Azure becoming a source of market confidence.
Compared to Meta, the market's reaction to Microsoft's earnings report was significantly more positive. Analysts believe Microsoft's greatest advantage lies in its proven ability to generate revenue growth from AI investments.
In its fourth fiscal quarter, Microsoft’s Azure cloud business revenue grew 43% year-over-year, marking the fastest growth since early 2022 and exceeding analysts’ previous estimate of 40%. For the fiscal year ending in June, Azure’s annual revenue surpassed $100 billion for the first time.
Microsoft CFO Amy Hood said that demand for cloud services continues to outpace supply, and she expects Azure's growth rate to accelerate to approximately 45% this quarter.
Jefferies analyst Brent Thill called Azure's growth "impressive." Bernstein analyst Mark Moerdler and Morgan Stanley analyst Adam Wood also acknowledged Microsoft's performance this quarter.
Microsoft’s AI assistant business also shows signs of growth. Microsoft CEO Satya Nadella stated that the number of paid users of Microsoft 365 Copilot has exceeded 30 million, up from approximately 20 million three months ago.
Bloomberg Intelligence analyst Mandeep Singh believes Microsoft’s strategy of promoting AI assistants through its Office ecosystem is working.
Meanwhile, Microsoft has not continued to significantly increase its capital expenditure guidance, unlike some of its competitors.
Microsoft's capital expenditure for this quarter reached $41 billion, a 70% year-over-year increase, slightly below analysts' expectations of $42 billion. The company expects capital expenditure to reach approximately $175 billion in 2026, below the previous market expectation of $190 billion.
Hu De stated that the company reassessed the useful lives of its data centers and office facilities, concluding that these assets may continue to be used for approximately 10 years, resulting in some capital expenditures being reclassified as operating expenses on the accounting statements.
Microsoft stated that, aside from this accounting adjustment, its investment plan for 2026 remains unchanged. The market believes that while Microsoft’s capital expenditures remain substantial, growth in Azure and increased adoption of Copilot demonstrate that these investments are generating commercial returns.
The AI competition has entered the validation phase, and the market is beginning to distinguish between input and output.
Both Microsoft and Meta are betting on AI infrastructure, but investors are using different criteria to evaluate the two companies.
Microsoft is demonstrating how AI is driving cloud business growth, with increased Azure revenue, rising Copilot user adoption, and stable capital expenditure expectations making the market more receptive to its investment thesis.
Meta still needs to answer a key question: When will its massive investments in AI infrastructure truly change consumer behavior and generate new revenue streams?
After AI competition moved from technological development into the commercialization phase, Wall Street’s focus has shifted from “who invests more” to “who can prove their investments are effective.” Microsoft is currently gaining this proof, while Meta remains in the waiting-to-be-validated stage.