Diverging Investor Reactions to Aggressive AI Capex at Meta and Microsoft

Meta and Microsoft drew investors’ attention in their earnings releases after the market close on January 28, as they outlined more aggressive AI data center investment plans. Meta (META) reported fourth-quarter 2025 advertising revenue and guidance that beat analyst expectations, while Microsoft (MSFT) delivered cloud results that were in line with or slightly below consensus. Immediately after the nearly simultaneous earnings announcements, Meta’s share price jumped more than 6%, whereas Microsoft’s fell by around 7%.

Morgan Stanley analyst Keith Weiss pointed out that while Microsoft’s capital expenditures (CAPEX) rose 66%, revenue growth in its Azure cloud business was only 38%. He relayed that investors are concerned the cloud division’s growth may be insufficient relative to the massive capex. In response, Microsoft CFO Amy Hood explained that the company had allocated some of the new AI computing resources to internal development teams working on its own product, Copilot, and that cloud growth would have been higher if all of those resources had been dedicated to Azure.

On the surface, the CFO’s explanation sounds like a reasonable justification for revenue growth lagging capex growth. But on closer inspection, her remarks contain an unintended admission that Microsoft’s Copilot product is less competitive than rival AI offerings.

In the company’s earnings presentation, Microsoft disclosed that revenue in the Microsoft 365 commercial segment grew 17% in total, driven by higher revenue per user from existing 365 commercial customers adopting Copilot and a 6% increase in user count. Stripping out the user growth, however, it suggests that the share of existing customers that have actually switched to Copilot remains relatively low.

In theory, Microsoft, with what is arguably the world’s most valuable portfolio of enterprise SaaS products, is in an ideal position to upsell by attaching AI add-ons to its existing software. In reality, though, its installed base has been lukewarm toward Microsoft’s AI products. This lines up with the recent trend of deriding Microsoft as “Microslop,” a play on words that mocks the company.

On major U.S. online communities, a new term, “AI slop,” has emerged as a catch-all label for low-quality, clickbait content generated by generative AI. Late last year, Microsoft CEO Satya Nadella wrote in a blog post that calling AI-generated content “slop,” a disparaging term, is not constructive for the future. That remark enraged users who had grown weary of years of incessant promotional pop-ups across Microsoft products, including the Windows operating system, pushing them to use Copilot. In response, they went so far as to stick “slop” onto the company’s very name.

The company has integrated its Copilot product, built on OpenAI’s GPT models, into nearly all of its existing software offerings. In doing so, however, it appears to have prioritized boosting AI adoption rates over considering the actual user experience. By focusing on AI integration for its own sake, Microsoft has failed to develop products that meaningfully enhance users’ productivity.

Meta had rattled investors when it set up a new AI research team and doubled down on AI data center investments after the performance of its latest open-source generative AI model, Llama 4, fell short of expectations. Daily active users have since inched up, and, more importantly, revenue per user rose 16% year over year, driving the company’s results. Meta CFO Susan Li explained on the earnings call that the company’s aggressive AI investments have led to better content recommendation algorithms and ad targeting, which in turn have kept users on Meta’s platforms longer and contributed to higher advertising revenue.

In fact, Meta’s massive AI investment began back in 2021, before the advent of ChatGPT. That year, Apple introduced an “Ask App Not to Track” option for iPhone users with the release of iOS 14.5. As a result, social media companies such as Meta and Snapchat (SNAP), which had relied on collecting behavioral data to deliver personalized, targeted ads, saw their ad performance deteriorate and ad prices plunge. When this started to weigh on earnings, founder Mark Zuckerberg began investing in GPUs as the answer.

In early 2023, in an article titled “Why Snapchat Delivered Disappointing Earnings: The Answer Is AI,” I wrote that the performance gap between companies with AI capabilities and those without would only widen.

Unlike Snapchat, Microsoft is a giant enterprise that generates massive B2B software revenue and profits, so it is not at risk of collapsing. But it will need to refocus on the idea that the essence is not AI itself, but user experience and problem-solving, in order to regain investors’ support.