Wall Street keeps falling for AI hype

Investors are a fickle bunch. They just need to hear a whisper of a rumor, or spy a flash of a headline crossing their Bloomberg terminals, and off they go on a hard sell-off, sure that whatever they heard is going to blow up their positions, when it's just as likely it's their reaction doing the damage.

We have seen this play out over and over, where one of the big AI labs makes an announcement, or even strategically leaks a rumor, and it sends established tech stocks plummeting. It's important to remember AI announcements are often full of PR bluster, making it easy for people outside the tech bubble to blow them out of proportion, and investors for some reason appear to be particularly vulnerable to this. 

As someone who regularly pushes back against the conventional AI narrative, I find it particularly frustrating that people who should know better accept what they're being told at face value without questioning it.

Perhaps the best example of investor overreaction was earlier this month when Anthropic announced Mythos, an AI model that can autonomously find and exploit software vulnerabilities. As I wrote in last week's column, it was a perfect combination of overexuberance and gloom playing out in one announcement. It was going to end all software bugs; it was going to help create super bugs at scale. 

For Wall Street though, it was another example of seeing an announcement, hearing the hype around it and overreacting in a big way by punishing all cybersecurity stocks. It's ridiculous on its face to suggest that this announcement is a major threat to established cybersecurity companies at this juncture, and probably ever, but that didn't prevent a swift sell-off.

Falling for Mythos myths

In fact, in the days following the Mythos announcement, stocks like CrowdStrike, Palo Alto Networks and Zscaler got hit hard on the news, with stocks already suffering this year, down around another 5-8% depending on the company. 

Mythos may end up being all it's hyped up to be, and still not be a substantive threat to these companies, which do much more than root out software vulnerabilities. They are full‑service cybersecurity platforms, and a tool from Anthropic, a company that has almost no security track record, is going to struggle to compete with vendors that have spent decades stopping attacks across networks, endpoints, identities and cloud environments. 

And it's not as though these vendors are just sitting still when it comes to agents and generative AI. They're building and buying fast and furiously to provide protections for the new kinds of security problems that AI brings. In fact, Anthropic itself chose CrowdStrike and Palo Alto Networks as founding partners in Project Glasswing, its program to deploy Mythos, a pretty clear signal that even Anthropic doesn't see its model as a replacement for these companies.

Just a week earlier, Anthropic was dealing with a major leak of its own involving large swaths of Claude Code's source code. That’s a colossal real-world mistake that will very likely lead to further security failures down the road, yet it barely registered compared to the reaction to a new product announcement that hardly anyone has used yet. Reports suggest that the same dynamic is showing up in private markets with investors reportedly offering Anthropic an astronomical $800 billion valuation in its next round of funding. 

Dragging down entire sectors

This isn’t just about one announcement or one company. It’s shaping how investors are thinking about entire sectors. We have seen many other examples play out in similar fashion since the dawn of the generative AI era with investors acting like sheep simply reacting and selling (or buying) without thinking through how much is hype and how much is real. 

It's not exactly a secret that SaaS stocks have taken a big hit over the last 18 months over fears that a combination of agents and vibe coding are going to somehow drive large, established enterprise vendors out of business. You would think the analysts on Wall Street would learn to take a more nuanced view on these announcements instead of jumping at every press release. 

I've written about this a lot (here, here and here), and my point is clear: the large SaaS vendors, for all their faults, do things that it's not possible for a single company to do on its own, and frankly, in most cases wouldn't want to because they have a business to run.

Look, I get that investors have to always be looking forward and searching for any news that could potentially impact their investments. I also understand that just because a company is on solid footing today, is not a guarantee of future returns. Nor can established companies simply rest on past successes. They have to be constantly on alert for disruptive forces in the market and change with the times. 

But it doesn't help anyone to be knee-jerk selling every time the big AI labs report a new feature. Investors need to stop treating announcements as outcomes and start looking for real evidence of disruptive change.
~Ron