Free cash flow. It’s all about free cash flow. Or, more to the point, lack thereof.

In “Tech Got No Love This Earnings Season,” I noted that EPS beats were wholly insufficient to drive outperformance for TMT stocks following earnings this reporting season.

Indeed, if you read that article (and yes, I realize that by telling you this, I’m virtually guaranteeing you won’t read it, but c’est la vie) you’ll discover that Comms Services and Info Tech actually underperformed the market post-EPS beats.

Why? Simple: Nobody cares about EPS right now when it comes to tech — or “tech,” in scare quotes, to account for the fact that when we talk about tech, we’re actually talking about a lot more than what’s officially classified as such.

Rather, people care about the interplay between capex and free cash flow. As discussed in “Have The Hyper-Scalers Lost Their Minds?“, “The Leveraged Buildout” and “The Problem, In A Nutshell…”, we’re in a weird space right now where it isn’t clear what investors actually want in terms of capex.

Ostensibly, they want to see some evidence that capital discipline’s still “a thing” in hyper-scaler boardrooms. I say “ostensibly” because it’s entirely possible that the first capex cut from one of those boardrooms will be met with a steep selloff if investors view it as confirmation that AI investments aren’t paying off fast enough, big enough or both.

That risk, in my view, will make management very reluctant to tip an actual cut to previously-announced spending plans. I think they’re pot committed to those plans, which in turn means they’re condemned (if that’s the right word) to another four quarters (at least) of worsening FCF burn.

The figure above shows you the quarterly breakdown for hyper-scaler FCF.

The problem — or potential problem — is that investors might’ve just run out of patience with that cash burn, leaving the hyper-scalers with no good options. If you cut capex, you risk conveying something you didn’t necessarily mean to convey about the ROI on AI outlays.

If you don’t cut capex, you risk four more quarters — i.e., the time between now and when FCF’s expected to inflect — of investors selling the stock post-earnings, which’ll be especially frustrating for management to the extent it’ll feel like being punished merely for reporting a predetermined outcome.

With all of that in mind, consider the figures below from Morgan Stanley’s Mike Wilson.

Those charts give you an idea about how sensitive investors are becoming to the cash flow issue.

“While a similar pattern is evident for EPS, FCF surprises are driving larger next-day price reactions,” Wilson wrote. Equity investors “are no longer overlooking FCF shortfalls as they often did earlier in the cycle.”

Nor, by the way, are credit markets ignoring the FCF trend. The figure below shows you how the AI borrowing binge has high-grade tech bonds trading wide to the overall IG universe.

That really is remarkable. I’m not just posting charts to post them. During the original pandemic panic, IG tech credit traded 80bps rich to the broader USD IG universe. Now, the “same” tech bonds are 20bps wide.

Again (again, again), the issue is that if investor patience has run out, it happened way too early. Because it’s going to be a year, at least, before hyper-scaler FCF inflects.

Unless of course one of them decides to actually cut capex, at the considerable risk not just of torpedoing their own shares, but of tipping AI sentiment dominoes more generally.