Big Tech delivers strong earnings, but AI spending fears rattle investors

With recent layoffs and more capital spending, Meta is testing the boundaries of investor patience

Despite blockbuster revenues, rising AI investment costs and doubts over returns left investors in Microsoft and Meta less than impressed

Yesterday was ‘Big Tech Results Day,’ as Amazon, Google-parent Alphabet, Microsoft and Meta all reported quarterly results. Whilst all four results seemed positive, the after-hours markets delivered differing verdicts.

Alphabet posted a chunky 22% rise in revenues on the same period last year, largely driven by a huge jump in Google Cloud revenues which climbed 63% to $20bn – far ahead of analyst expectations.

Alphabet share prices subsequently rose a little over six per cent in after-hours trading.

AWS drove Amazon’s revenues to $181.5bn after recording its strongest quarterly growth since 2022, and Microsoft grew Azure by 40%, with its AI products now creating annual recurring revenues of $37bn.

Amazon shares increased 4% off the back of its earnings announcement but Microsoft’s stock dropped slightly – around 2%. The losses were recovered after-hours, but Microsoft stock ended the day pretty much as it started it.

Why was this case given the fact that also beat revenue forecasts and increased profits to $38 billion?

Trouble at OpenAI

The more muted reaction might well reflect some nervousness about the extent to which Microsoft’s spending on AI has started to affect its cashflow. Microsoft’s free cash flow fell almost $6 billion on the same quarter a year ago.

Microsoft’s stock is down about 11% so far this year, with many questioning the massive costs of its AI push and its colossal investment in OpenAI, with a recent report in The Wall Street Journal suggesting it is going to undershoot revenue and user targets.

The report quoted OpenAI CFO Sarah Friar saying that she harbours doubts over whether OpenAI is ready for IPO and has serious concerns about whether the company can meet its financial obligations to pay for future compute capacity if its revenue doesn’t grow fast enough.

The same report featured a comment that Friar also doubted whether OpenAI “was ready to meet the rigorous reporting standards required of a public company.”

This is all bad news for Microsoft and that may be reflected in the muted response to what looks like a strong set of results. (It’s even worse news for Oracle but that’s one for another day).

OpenAI called the WSJ report “clickbait” – which wasn’t a denial.

Meta

The market reaction to Meta’s announcement that it had grown revenues by 33% to $56.3bn, beating expectations, was a reduction in its stock of more than five per cent in after-hours trading.

Why? The Meta earnings announcement was accompanied by an announcement of a big increase in capital investment. Meta expects to spend between $125bn and $145bn on capital projects this year, around $10bn more than it had previously planned.

Up until recently, announcements of redundancies and increases in capital expenditures in expectation of revenues from AI would boost earnings. That pattern now seems to be, if not breaking down, certainly less predictable.

LLMs gotta make rent

Meta’s problem is that, unlike the other hyperscalers, it doesn’t have a cloud business with millions of enterprise customers to generate revenue. Its revenues come from largely from advertising, and investors seem to be losing patience with Mark Zuckerberg’s grandiose statements about AGI, especially given that the company burnt billions gambling on the metaverse.

Jim Piazza, Chief AI Officer at Ensono - and former Meta infrastructure lead summed up the challenge:

"Meta beat the quarter - and beat it well - but the result almost sidesteps the bigger debate. Investor attention is fixed on rising capex spend, and a strong quarter doesn't close that conversation.

“What we're seeing across the hyperscalers is AI infrastructure demand being subsidised by workforce reductions, and that trade-off isn't going away. Datacentre land purchases are drawing more scrutiny too, as rising energy costs make the scale of rollout harder to justify.”

The scale of that rollout is vast. The Big Four are on course to spend around $650bn building out AI data centres and infrastructure this year. Piazza continued:

“Meta has one of the cleaner AI monetisation stories in this group because it can point AI directly at its core ad machine: better targeting, better creative automation, better ranking, better engagement, higher advertiser ROI. That's a much easier case to make than 'trust us, someday the chatbot will pay rent.' “

The chatbot paying rent is clearly on minds at both Microsoft and Anthropic. All Github CoPilot plans are moving to usage-based pricing from the beginning of June and Anthropic is also tinkering with its pricing tiers, removing Claude Code from the $20 a month pro tier for new subscribers and removing the ability for Claude Pro and Max subscribers to use their flat-rate subscription limits with third-party AI agents like OpenClaw.