Checking in on the SaaS Bear Case(s)
Dispatches from another red day
Bit of a different article today, just going to rip this one off the cuff.
Another day, another Anthropic product release, another washout for software. It sure feels like Groundhog Day for software investors. My phone lighting up with my SaaSpocalypse watch list clearing new 52-week lows every 5 minutes.
There’s something that’s a bit surreal about all of this, as the much-feared negative impact of AI on software hasn’t even materialized yet for many of these businesses who are down 50-70%. This poses serious reflexivity risk, that the very fear of these businesses ends up destroying them before AI is able to. If your most agentic (heh) people can leave with their equity comp down 70% to join an AI startup, your company is left in a difficult situation when it comes to building for the AI future.
It feels like another lifetime ago (eight months) when I posted a list of headwinds for software in the AI era
The commenters were more than happy to share a few more with me that I had missed:
Being a System of Record isn’t an advantage anymore because agents lower switching costs.
Agents will do the work making UI/UX irrelevant, further commoditizing software.
AI will be a deflationary force so overpriced subscription model is dead, especially when Claude and ChatGPT cost $20/mo.
Buy v Build framework has flipped to build for many point solutions.
Outcome pricing is subjective and difficult to verify, so many seat-based companies will fail in the transition.
I could go on, but you get the idea. Bear cases are stacking up by the day. You know what isn’t appearing? The bull case. Where is the category leading agent solution from an application software incumbent? Beyond, like, Palantir, who is blowing the doors off by capturing AI budget and accelerating?
Since I posted that in August, IGV is down 30% and SMH is up 42%.
This chart was going around Twitter today. The biggest 2-day drop on record for Software vs Semis, so as extreme as this trade has gotten, it’s accelerating.
And honestly? It might be as simple as “nobody wants to own stocks that can go down 5% every day for the foreseeable future for no reason.”
Here’s a few things guiding my thinking when it comes to software. I think a bunch of other folks are seeing these things, which is causing a real loss of confidence in incumbent software:
The underlying models keep getting better.
More and more of my department’s time and workflows are being sucked into Claude and ChatGPT. Everything from coding to analytics to writing. When I step back and think about it, I’ve never actually experienced a paradigm shift in my workflow that’s been this extreme.
Vibe coding is getting way, way, better. It honestly sucked six months ago and now I find myself constantly surprised on the internal tooling we’re able to stand up using vibe-coded solutions. Sure we’ll replace it as it scales, but it works for now. Or, vibe coding will keep getting better and we won’t.
Executives are getting way, way more AI-pilled. This is pushing adoption deeper into the Enterprise. It is making the seat-compression bear case feel more real. It will likely make contract renewals more fierce.
Every single AI use case we are evaluating is not being won by incumbents. A couple expert calls and check-ins will validate this for any investor. It’s damning.
I am talking to my friends at AI startups and these folks are a bunch of lunatics. They are grinding so hard. It is difficult to articulate. They see the opportunity and are going in for the kill. I am talking people with young families grinding to the bone 6-7 days a week. My friends at incumbent software companies are talking about how challenging it is to drive culture change. This is a big deal and a real problem.
There is always a risk with these sorts of articles that you bottom-tick it. And maybe that will be the case here. But things feel awfully slippery, and there’s not much to catch your fall below when it comes to valuation for these names. Good luck out there to all.






Love this format!
Great post. “Every single AI use case we are evaluating is not being won by incumbents.” Can you give a few examples?