CIO unpacks earnings from four of the mag seven
Wednesday was a test for technology. Earnings reports from four of the ‘magnificent seven’ companies, Meta, Amazon, Microsoft, and Alphabet, were heralded as a great test for the AI mega trend. After years of rewarding visionary leaps in technology and vast capital expenditures to build AI infrastructure, Q1 earnings were seen as a moment when the market’s metrics for success would turn, from AI promises to AI revenues.
The results lived up to those expectations. All four of the big tech names that reported on Wednesday beat earnings estimates, but not all four enjoyed the same performance. Alphabet showed real revenue growth from AI-enabled cloud services as well as other AI-connected lines of business. Microsoft and Amazon gave more mixed pictures of AI monetization against higher capital expenditures. Meta promised a vision more than it showed AI revenues. Markets seem focused, for now, on revenue over vision.
“The story from last night is ‘show me the money.’ That is the bottom line. Meta got punished for offering a vision rather than offering revenues. Google got rewarded for showing revenues,” explained Elliot Johnson, CIO at Evolve ETFs. “These companies that we’re talking about all have lots of other products and services. It’s very clear from yesterday those products and services are thriving. But capex is up big time… The question that people have had on their minds, and I’ve been hearing this for the past six to eight months, has been: ‘is the spend justified by revenues.’”
Some of the shift in investor outlooks on AI, Johnson explains, has come from the rise of Anthropic as one of the fastest growing companies in history as measured by paying user growth. Revenue for the AI company grew from $1 billion at the start of 2025 to $19 billion in February of 2026. Much of that growth has come from a focus on enterprise clients and serving those businesses through the creation of agentic AI that’s meant to function as a coworker. Johnson contrasts this with OpenAI’s focus on personal use, which has resulted in huge popularity but less revenue growth.
Now the hyperscaler tech companies, which can sometimes be used as proxies for private AI names like Anthropic and OpenAI, are being tested along those same metrics. That’s why Google’s 63 per cent increase in cloud revenue was so rewarded by investors. Cloud businesses, it seems, are some of the first major sources of AI revenue for the hyperscalers. Meta, conversely, lacks a cloud business and therefore doesn’t have that means of turning its AI usage into real revenue.
Meta, in Wednesday’s earnings report and in many of its AI capabilities, has held to the old silicon valley mantra that worked in past tech iterations: that building a ‘toy’ to attract users should be the priority, with revenues due to follow. Johnson notes that while investors had previously rewarded the impressiveness of those ‘toys,’ there’s been something of a maturation in the market that comes with any promising technology. Once some players in that tech show they’ve found real revenue generation from their technology, they can reassess pricing based on which products will have continued success and which products might not. Mark Zuckerberg’s promise to provide a superintelligence Meta’s audience of one billion people is not greeted as warmly as significant cloud revenue growth or Claude’s tangible capacity to build a slide deck.
“I’m thinking about this in terms of real tangible things in terms of the tasks, the activities, the achievements of this technology and what I’m spending on it,” Johnson says, describing his mindset. “And I think this is a very natural place to get to, but it does mean that it’s all about ‘show me the money.’”
For all the ways Meta may have stumbled in this quarterly earnings report, Johnson says there’s still a bull case to be made for that company, if only because they still have one billion users across their platforms. He describes the magnificent seven as “money printing companies,” capable of weathering negative investor reactions.
The massive revenues that these companies can still show are also key to their recent spate of debt influences to finance capital expenditure on AI infrastructure. While the revenue tests for these companies may get harder if they issue too much debt, he believes that the existing paths to monetization for AI infrastructure can continue to justify these debt issuances and reassure investors on both equity and credit markets that this debt is sustainable.
Underlying all these investor tests and revenue beats is an ongoing AI infrastructural buildout that requires vast quantities of hardware: semiconductors, RAM, CPUs, copper wire, natural gas, the list goes on and on. With AI capacity looking something like a highway, where new lanes and new capacity simply induce demand until traffic is at a new peak, Johnson is instructing his team to look for areas in this market that enjoy seemingly infinite demand.
“I’ve been saying this a lot to the portfolio team here at Evolve recently that investors should just be owning companies that have essentially infinite demand. TSMC is a perfect example. Don’t overthink it. You have a company that has essentially infinite demand. There is a lineup out the door that is infinity long just wanting to buy their chips. Don’t overthink it. Go own that stock,” Johnson says.
