Inside Goldman’s AI Winners and Losers List in Software and SaaS
The rational investor's guide to the 26 winners, the 41 losers, and 3 thoughts on the SaaS sell off.
Goldman Sachs published a report saying AI will incinerate 41 software companies and spare 26 others. They didn’t publish the names and I am not cool enough to be invited into their research platform. So I tracked them down, disagreed with half of what I found, and then used AI to build a full report. The irony of using Claude to analyze whether Claude is taking everyone’s jobs and cause a 2028 depression was not lost on me.
Do I think Goldman is right? Partially. The selloff was real, the math behind it seems sound, and some of these companies are genuinely in trouble. But indiscriminate fear creates indiscriminate mispricing, and when $2 trillion comes off the board in 18 months, the good businesses get marked down alongside the bad ones.
What follows are my three lessons from the selloff. And behind the paywall is the full report: all 67 names, the Goldman Winners and Losers baskets, technical screens, and my own take on which companies are genuinely threatened, which aren’t, and which are being priced as if the outcome is already known when it isn’t. Of the names on the list I own only GOOG & CPAY.
And yes, I built the report using the very technology I'm questioning, which either makes this more credible or completely disqualifies me, and I'm genuinely not sure which.
Lesson One: Valuation Is a Claim on Distant Cash Flows
The SaaS selloff was arithmetic and only a matter of time in my opinion.
Forward P/E multiples peaked near 50x in 2024 and compressed to roughly 20x in under 18 months. Roughly $2 trillion in market cap came off the board. And from inside a SaaS portfolio I have no doubt it felt extreme, but from the outside looking in, it was just math.
Remember that when you pay 50x forward earnings, you are betting that a company’s earnings will be large, predictable, and growing for 10 to 15 years. Maybe even longer. That duration makes the valuation extremely sensitive to small changes in assumptions. Even a 2-3% shift in a long-duration discount rate can move a valuation 20 to 30% with no change in near-term fundamentals. SaaS multiples were expensive going into this and on a backwards looking basis the growth rates had earned the premium. What the premium did not include was any buffer for doubt.
AI and our boy Claude introduced doubt. Not certainty of disruption, just genuine uncertainty about whether the long-run cash flow projections would hold, which was and is enough for irrational humans and hedge funds to hit sell like a maniac. You see, when terminal value comes into question on a long-duration asset, the market does not wait for resolution, it reprices first. The selloff was a correction that needed a catalyst and AI provided one.
Lesson Two: You Cannot Have It Both Ways
Two narratives are running in parallel right now and they are in direct conflict.
