Software stocks are getting crushed. Not selectively. Not tactically. Almost indiscriminately!
If you scan the sector today, the picture looks less like a rotation and more like a liquidation, maybe even capitulation – growth leaders, category winners, infrastructure platforms, horizontal SaaS, vertical SaaS, all bleeding together.
The chart below I came across on X today captures this brutally well: across dozens of well-known software companies, drawdowns from recent highs cluster around levels that would normally be associated with existential crises rather than cyclical uncertainty.
It looks like when it comes to investing, “Hardware” is the new “Software.” - Paul Andreola
This feels like a collective repricing of what software is worth in a world where AI agents can increasingly do what software used to do, and where the durability of cash flows is questioned. Maybe you have visibility into the next two years, but what about 2030 and beyond? Can you have confidence in the cash flow production capabilities of most SaaS names?
From a stock picker’s perspective, this is paradoxically close to an ideal environment. What more could investors hope for? An above-average sector – which I believe “software” is – being sold with almost no discrimination between fragile business models and genuinely durable ones?
This is precisely the kind of regime in which careful company-specific analysis can matter again. When capital exits an entire category because it feels “too hard,” or because of “career risk” because one can’t see light at the end of the tunnel yet, the opportunity is rarely that everything is broken.
More often, it is that the market has stopped distinguishing between fundamentally different types of businesses.
I recently came across a tweet that captured this mood perfectly. The argument was that what we are watching is not a normal drawdown but a loss of confidence in an entire business model by investors who cannot afford to be wrong on a one-year horizon. When uncertainty rises, nuance disappears. Portfolios get simplified. Complex theses get abandoned. Software becomes a single trade rather than a spectrum of economics.
At the same time, the competitive boundaries of software are visibly shifting. Robinhood, historically a brokerage platform, is now experimenting with full-service tax filing, estate planning, and dedicated financial advisors – bundled into a single app experience. This is not just product expansion. It is a signal that the traditional borders between software categories, financial services, and professional services are eroding.
When software becomes cheaper to build and easier to distribute, adjacent industries start to collide. The question for investors is no longer whether disruption will happen, but where it will hit first and how deeply it will cut into existing profit pools.
This is why the very recent episode of the Invest Like the Best podcast with guest Gokul Rajaram felt unusually important to me. I’ve summarized key takeaway of many episodes from the show – the most recent one being the Henry Ellenbogen show – so let me once again give a major shoutout to Patrick O’Shaughnessy for the incredible value he creates by sharing these episodes freely.
Learning from Henry Ellenbogen
A few days ago, I came across the fairly new Invest Like The Best episode featuring Henry Ellenbogen, founder of Durable Capital Partners.
Rajaram has built products at Google, Facebook, Square, and DoorDash and invested in more than 700 companies. In the first half of the conversation, he does not offer a generic “AI will change everything” narrative. Instead, he implicitly lays out a set of mental models for distinguishing between software businesses that are structurally vulnerable and those that are surprisingly resilient.
When you connect his ideas about systems of record, pricing models, data durability, workflow depth, and stickiness, something close to a coherent framework emerges – one that can help investors think more precisely about which software stocks are most prone to disruption and which ones are quietly protected by deeper moats than the market currently acknowledges.
In what follows, I’ll try to reconstruct and extend that framework. Not as a checklist, but as a way of thinking about software economics and durability in the age of AI agents – and, ultimately, as a lens for navigating what might be the most indiscriminate sell-off the sector has seen in years.
The Wrong Question to Ask
If you listen to most conversations about AI and software, they tend to revolve around a binary question:
Will AI disrupt software companies or not?
That framing is seductive because it’s simple, but it’s also deeply misleading. The question is not whether disruption happens. It is where it happens, how fast it happens, and which layers of the software stack are structurally exposed versus structurally insulated.
Gokul Rajaram presented a couple of angles to look at software businesses to assess their vulnerability – or defensibility:
The rest of this post is FREE to read. I only ask that you subscribe, which costs nothing and can be undone with a single click whenever you like.
The wall also helps keep my writing from being freely scraped by AI crawlers, so I hope you understand the small extra step.
In return, you get every deep dive, earnings review, and process piece I publish, straight to your inbox.




