On February 11th, the IGV software ETF dropped another 3%. As of yesterday, February 25, it was at or about the same level. A bunch of once-beloved SaaS stocks were already down 50% or more from their highs. So when Bloomberg’s Odd Lots kicked off,1 the mood felt half-joking, half-uneasy.
“It has been ugly in software. Everyone's throwing around the term SaaS-pocalypse,” Tracy Alloway says. Joe Weisenthal adds that Salesforce “has basically been cut in half… since its peak.”
They laugh, though uneasily. The market isn’t.
At the centre of the episode was a simple question that’s actually not simple at all: Is AI about to wipe out software companies — or is it just forcing them to evolve?
To dig in, they bring on Jared Sleeper, partner at Avenir, an investor who’s seen software from almost every angle — early-stage startups, growth investing, public markets. His takeaway? The fear feels intense. The numbers don’t — at least not yet.
This isn’t a story about collapsing revenue. It’s a story about collapsing certainty.
The Rorschach Problem in Software
Sleeper makes a point that lands hard: most investors don’t really understand how software works.
“There’s very few folks, even on the buy side, who really understand how software works. It’s one of those Rorschach test kind of sectors.”
If you invest in Lululemon, you can walk into a store and try on the product. If you own Coca-Cola, you know what Coke tastes like. But most investors haven’t configured Salesforce. They’ve never migrated an ERP system. They experience software as numbers on a dashboard, not as the messy infrastructure running inside a company.
That gap matters. When markets get nervous, people retreat to what they understand. Software’s complexity makes it easy to fear — and easy to misprice.
The Numbers Aren’t Terrible. The Fear Is.
Here’s the strange part: the fundamentals haven’t imploded.
Growth peaked at over 40% in 2021 during the pandemic surge. It’s now closer to 18%. Net retention sits around 110%. Some companies are even guiding toward acceleration.
As Sleeper puts it, “It’s very hard to argue that this is something that’s happening today and showing up in the numbers.”
So why are stocks getting crushed?
Because markets don’t just price the present. They price the future — especially the risk that the future looks nothing like the past.
Sleeper frames it cleanly:
“There are two arguments against software right now. One is the world is going to stay the same, but software is just going to get a lot cheaper… The second is the world's about to get really weird.”
That second idea — the weird one — is driving the selloff.
We’ve seen this movie before. When ChatGPT launched, Chegg’s stock collapsed almost instantly. Revenue didn’t fall first. The market moved first. Investors sensed the threat.
Now the question is: who else is Chegg?
Code Is Cheaper. But Software Is Bigger Than Code.
Joe pushes back on the doomsday logic: if AI makes writing software cheaper, isn’t that actually good for software companies?
“If you're Salesforce and you can trivially reduce the cost of building software, that's also a benefit for you.”
Sleeper agrees the tech works. “It’s producing working code.” But he’s clear that software isn’t just code.
When companies buy enterprise software, they’re paying for more than functionality. They’re buying:
- Ongoing support
- Integrations with everything else they use
- Regulatory compliance
- Brand trust
- Standardization across their workforce
Sleeper introduces a term for this: herd familiarity.
“Everyone on earth knows how to use your software… that simplifies the training and onboarding workflow.”
Zoom wins partly because everyone already knows how to use Zoom. That convenience is worth money. Software becomes less about raw capability and more about coordination.
The Big Shift: From Seat Pricing to Outcome Pricing
The real transformation might not be in code generation. It might be in pricing.
For years, SaaS charged per seat — say $1,000 per year for a salesperson who earns $250,000.
AI changes the framing.
Sleeper explains: “You used to think of us as something you paid $20, $30, $40 per seat per month for… Now we're just selling you an employee or the results of an employee.”
Instead of charging for access to a tool, companies could charge per result:
- Per customer support ticket resolved
- Per sales action completed
- Per workflow automated
That’s a completely different ceiling. If software replaces labor directly, pricing power expands — dramatically.
The bullish case isn’t thinner margins. It’s a much bigger revenue pool.
But What If the Model Makers Move Up the Stack?
There’s a catch.
If startups can resell AI intelligence at fat margins, why wouldn’t OpenAI or Anthropic just do it themselves?
Sleeper doesn’t dodge this.
“They have been racing towards the application layer.”
Claude Code. Codex. These aren’t just APIs. They’re products people download and use. That’s a strategic move. Control the interface, and you control the customer.
It echoes the cloud wars, where infrastructure providers slowly crept up into application territory. Only this time, it’s happening faster.
Data Isn’t Enough. Context Is.
Tracy asks about data. Isn’t proprietary data the moat?
Sleeper complicates that idea.
AI models might have incredible intelligence. But they still need context. What happened at the client dinner? What nuance isn’t in the CRM?
“You don't know what happened at the sales dinner last night unless the rep took really detailed notes. And… they do not take very detailed notes.”
Enterprise memory lives partly in software — and partly in human heads. The next battleground might be whoever captures that context most effectively.
The Quiet Problem: Profitability
Underneath the AI debate is a less glamorous issue: profitability.
Many software companies lean heavily on stock-based compensation. On a non-GAAP basis, margins look solid. On a GAAP basis, not so much.
Sleeper is blunt: “The median public software company has a 5% GAAP net income margin.”
Five percent. That’s thin.
Without meaningful GAAP earnings, there’s no real valuation floor. No dividends. Limited buybacks. When growth narratives wobble, there’s not much underneath to cushion the fall.
Layoffs and Adaptation
Sleeper expects companies to respond — possibly with layoffs.
“I do think we will [see layoffs].”
But he frames the shift more as evolution than elimination. He compares it to civil engineering. Engineers used to do calculations by hand. Now software does the math. Engineers manage the system.
“Each individual software engineer is way more productive than they were before.”
AI doesn’t necessarily erase roles. It changes them. At least for now.
Markets, Fear, and Short-Termism
There’s also a structural piece. Hedge funds and pod shops can’t afford to sit through big drawdowns. If risk spikes, they cut exposure.
As Sleeper says, “They have no idea when it will stop… You’re one OpenAI model release away from more fear.”
In a world where a single model update can move billions in market cap, caution becomes contagious.
So… How Weird Is This Going to Get?
Near the end, Joe sums up the unease:
“Getting from like here to there where, okay, AI really changes the nature of software buying does feel like you have to get into… weird territory.”
Tracy replies, “If you bet on weirdness… that’s a pretty good bet.”
That’s the real story.
Short-term performance looks fine. Long-term structure feels unstable. Markets are pricing in possibility — not proof.
Collapse, Compression, or Renaissance?
Sleeper’s framework leaves us with three paths:
- Software gets cheaper, margins compress, but the world mostly stays the same.
- Certain niches collapse Chegg-style.
- AI supercharges software, expanding its reach into labor itself.
Right now, the market doesn’t know which future we’re heading toward.
So it’s repricing the uncertainty.
As Sleeper reminds us, “Investors are sharp… They’re constantly looking.”
What’s clear is this: the software boom of 2021 is over. What replaces it could be contraction.
Or it could be something much bigger.
We just don’t know yet.
Footnotes:
1 Podcasts, Bloomberg. "Jared Sleeper on Which Software Companies Will Survive the SaaSpocalypse | Odd Lots." YouTube, 19 Feb. 2026
