Why AI Software Stocks Are Falling Despite Strong Earnings

**Fexingo** (0:01)
If today's tech conversation gave you something usable, great, because the market is giving AI software stocks nothing but pain this week.

**Luna** (0:10)
Yeah, I've been watching this. Adobe down 16% in five days, Salesforce off nearly nine, ServiceNow down more than 10
And these companies just reported earnings that by any normal standard were fine.

**Fexingo** (0:24)
Right, so what gives? I think there's a specific dynamic here that's both more interesting and more worrying than just tech stocks are getting crushed. And before we go deep, quick, honest thing, a handful of listeners chip in monthly through by me, a coffee.com/fexingo.
And that's literally what funds making episodes like this. No ads, no sponsors trying to shape the angle.

**Luna** (0:49)
Yeah, it's genuinely listener supported. And we try to make it worth your time.

**Fexingo** (0:54)
So let's talk about what's actually happening to AI software stocks. Because the narrative in April was that AI would save enterprise software margins forever.
And now?

**Luna** (1:05)
Now it feels like the market is suddenly asking, okay, but where's the moat? If every company can bolt on an Elon, what's proprietary about Salesforce Einstein or Adobe Firefly?

**Fexingo** (1:18)
Exactly. And that's the heart of it. Let me put some numbers on this.
According to data from a major enterprise AI spending survey, we cited in episode 43, AI-pilled firms are spending about $7,500 per employee per month on AI tools. That's enormous. But the market is realizing that a lot of that spend flows to the model providers, OpenAI, Anthropic, Google, not to the software layer.

**Luna** (1:44)
So the software companies are essentially reselling someone else's intelligence with a thin layer of workflow on top.

**Fexingo** (1:52)
Right. And when inference costs are dropping as fast as they are, we talked about this in episode 41, inference costs are crashing faster than training costs, the value shifts downstream. But the software companies aren't capturing that value. Their margins are actually under pressure because they have to pay for API calls.

**Luna** (2:11)
Let's use Adobe as a concrete example. They reported fiscal Q2 earnings on June 11th. Revenue was up 11 percent year over year.
Digital media net new annual recurring revenue came in at 487 million, above the street's 462 million estimate. Earnings per share beat by a few cents. And the stock dropped 8 percent the next day, then another eight over the week.

**Fexingo** (2:38)
That's the pattern.
Strong quarter, stock gets punished. Why? The whispers on the street are about guidance. Adobe's Q3 revenue forecast came in at about 5.37 to 5.42 billion, and the midpoint was roughly in line. But the market wanted more. Specifically, they wanted evidence that AI features like Firefly are actually driving incremental subscription revenue, not just being bundled in for free.

**Luna** (3:05)
And Adobe's not alone.
Salesforce reported in late May revenue up 11 percent, earnings per share beat by 8 cents, but they lowered the top end of their full year revenue guidance. Stock dropped 16 percent in a single day.

**Fexingo** (3:21)
16 percent on a beat. That is not a valuation story. That's a narrative break. The market is repricing these companies as if the AI premium is gone.
Look at the IGV, the iShares expanded tech software ETF, down 5.2 percent in the last five days. That's an index of over a hundred software companies.

**Luna** (3:43)
So is this a rotation? Investors fleeing software for AI infrastructure? Because NVIDIUS down only 2 percent, AMD is up almost five, ARM is up 10

**Fexingo** (3:56)
Partially, but even infrastructure has split. Super Micro Computer is down 31 percent in five days. That's a company that makes servers for AI data centers. Their last quarter showed revenue growth of 30 percent year over year, which is solid, but gross margins compressed from over 15 percent to under 12
The market hates margin compression.

**Luna** (4:20)
So the story is not AI good, non-AI bad. It's AI infrastructure with pricing power good, AI infrastructure with thin margins bad, AI software with no moat very bad.

**Fexingo** (4:34)
That's exactly it.
The software companies are in a bind, because to keep growing, they need to show that AI features drive new seats or higher prices. But most of them are giving AI a way to defend their installed base. Microsoft is bundling Copilot into Office 365 with no price increase for many enterprise customers that sets a ceiling for everyone else.

**Luna** (5:00)
What about ServiceNow? They reported earnings in April, subscription revenue up 22%, remaining performance obligations up 30%.
Stock initially rallied, then gave it all back, and then some.

**Fexingo** (5:14)
ServiceNow is actually interesting because they have a more credible AI story.