SaaSpocalypse 2.0.
With Jez Rowe and Cam Lynch
SaaS didn't die in the February 2026 sell-off. It bounced back within about 10 weeks as earnings showed AI was helping software companies grow. The real pressure is cost: token bills, freemium plans and expensive frontier models. Jez and Cam land on proprietary models as the next move, and on the real question being who ends up with the margin.
What this episode answers
What was the SaaSpocalypse in 2026?
The SaaSpocalypse was the February 2026 sell-off in which software companies lost around $2 trillion in value on fears that people would build their own software with AI. The broader market was fine. Salesforce fell about 30% and Workday 33%. It lasted roughly 10 weeks. Software stocks then rose more than 40% off the low, to within about 13% of their highs, once earnings showed big SaaS companies growing 30 to 40% using AI.
Why is Canva more exposed to AI than Figma?
Canva is more exposed because it sells direct to consumers, who can now make simple designs straight in Claude or ChatGPT. Someone making a birthday card or wedding invite no longer needs a design platform. Figma is embedded in product teams at businesses, which is why it has kept growing. In August 2026, Canva cut its quarterly revenue growth forecast from about 30% to 20%. Direct to consumer software is the most exposed to people using AI directly.
Does the freemium model still work for AI software?
Freemium is under serious pressure once AI is in the product, because every free user burns tokens that cost the provider real money. The 2010s SaaS model, where your 10,000th customer cost you almost nothing, doesn't hold when each user runs AI. Canva shows the problem. The likely story behind its August 2026 forecast cut is that AI got used too much on free plans, not too little.
Can a SaaS company afford to run its product on frontier AI from OpenAI or Anthropic?
Not at scale, because frontier AI inside a SaaS product burns expensive tokens on every user. Cam Lynch, founder of Vibe CFO, puts it simply: proprietary models are the next move, purely for cost. Harvey, a legal AI company, runs enormous context and reasoning over legislation worldwide, and in August 2026 it moved toward its own model, as Rippling and Pinterest have. Owning the model also protects the IP, and better open source models make the switch easier.
Will AI kill SaaS companies?
No, AI won't kill SaaS companies, but it is squeezing their margins. The question was never whether SaaS survives, it is who ends up with the margin. The idea that businesses would build their own Xero or HubSpot with AI has proven wrong. Instead, frontier AI costs software companies a fortune per user, customers push back on price rises, and users build overlapping tools in their own Claude and Codex subscriptions. The SaaS companies that control their AI costs keep the margin.
Episode notes
In February, software lost two trillion dollars. Everyone said AI was coming for SaaS.
Ten weeks later it was over. The big platforms reported earnings growing 30 to 40%. Nobody built their own Xero.
So that's settled. But something else started.
Canva cut its growth forecast. Not because people stopped using the AI. Because they used it too much, on free plans. For fifteen years the 10,000th customer cost you nothing. That's what made freemium work. Tokens changed it.
And Harvey, the legal AI company, just shipped its own model rather than keep paying frontier prices on every legislation search.
Cam and Jez on what actually broke, why direct-to-consumer is more exposed than enterprise, and the question underneath all of it.
Not whether SaaS survives. Who ends up with the margin.
Five minutes. One thing that matters. We ignore the rest, so you can too.
Vibe CFO is Australian founded. An AI CFO for small and medium business, out of the box, governed and secure.
Vibe in 5.
Transcript
Read the full transcript
Ah, yet another big week in AI. This is The AI Exhale, hosted by Vibe CFO with Jez and Cam. Five minutes, one thing that matters, or maybe two. We ignore the rest so you can too.
Jez: Welcome to this week's news episode of the AI Exhale, and we're going to call this one the SaaSpocalypse 2.0. Early this year in Feb, we had software companies losing $2 trillion of value. And that's where you would have heard terms like the SaaSpocalypse.
So not the market, the market was fine. It was just software. Examples of Salesforce going down 30%, Workday down 33%. But then it came back more than 40% off that low, to within about 13% of its all-time high. So the whole extinction event lasted really only 10 weeks.
Cam: That initial SaaSpocalypse was around the thought that people are going to build these softwares themselves, which I think has proven to be wrong now. Like, that's not going to happen. Like, are people going to build a Xero or a HubSpot themselves? And then the flow on from that, what put that flame out, was that the big SaaS companies, following the Fed announcement, had started growing. Earnings season came in and they're growing at 30 or 40%. People are like, hang on. Using AI, these SaaS companies are actually growing their revenue. So we can't say that they're going to die when they're doing the complete opposite.
Jez: Revenue definitely maintained its speed and growth, maybe not forecast. So we want to talk about a couple of use cases here, around one of the darlings close to home with Canva. Just unpack that story a little bit.
Cam: So we look at what's happening with Canva. Canva's reduced its forecast. It had forecast revenue of about 30% growth for the quarter and it reduced that down to 20%, right? They might be seen as on the frontier of this, because Canva goes direct to consumer. So then you think, okay, what does that mean to go direct to consumer? Like, are they more exposed compared to enterprise? That's the argument, which I think is quite strong, is that Figma is actually used by loads of product teams.
And that's why they continue to grow, whereas Canva is threatened by people making birthday cards who just go straight to Claude or ChatGPT. You can imagine just building a birthday card in Claude or ChatGPT rather than using Canva. So I think that's their threat, that direct to consumer is exposed more to AI, like direct usage, direct building.
Jez: Yeah, no, I built our wedding invites on Claude. I didn't dabble with Canva. So this is really where the SaaS model, that nice business model that we saw through the 2010s where the 10,000th customer cost you nothing. Right. Of course, Canva with their freemium offering as a method of getting people on the platform, and they're using AI. Right. That's token cost.
Cam: It seems like this is part of the issue. I really question where freemium sits now, because of the cost to give the freemium model now with AI token usage. I think the story wasn't that nobody used AI in Canva. The story was that it got used too much as well. Like, it was just overused on these freemium plans and they had to go, oh, hang on a minute. We need to regroup here on how this is all done.
Jez: And hot off the press, Harvey, a legal AI company. Here we want to focus on the move to owning and using your own proprietary model and not depending on the now quite costly frontier models.
Cam: So Harvey is similar to Canva in the sense that you can imagine the amount of token usage they've got, with legislation and context, like context and reasoning looking over legislation. Not just Australian legislation, but across the world, it would be monstrous. So what they've worked out is it's economically unsustainable to use frontier AI like Anthropic and OpenAI inside your SaaS.
I don't think it's going to work. And that's why they're going to this proprietary model. I think this is the next move. Proprietary models with their context. Like, they build it, they're protected, it's their IP. I think it makes perfect sense. My takeaway on this is that SaaS companies are moving into building their own proprietary models, and frontier AI inside SaaS is not economically sustainable.
Jez: Some other examples being Rippling and Pinterest, who have made this move. But yeah, that legal case, nice in there. Just outside of cost, you've got those obviously digitised, dusty legal books with all those precedents. Like you say, they've got that wealth of information and knowledge to utilise.
If we tie the knot here, isn't this like the perfect storm to get to this conclusion? We've got a massive increase in open source models just getting really, really good. Kimi K3 was the first to fall, but there's been a couple more since we talked about that. Open source models just massively increasing in their use, their smarts. An explosion of SaaS usage with frontier AI, but costing the SaaS provider a fortune. Massive downward pressure from users who don't want the SaaS increased in pricing. And then they're playing around directly with their Claude and Codex subscriptions inside their plan to build some of these things that are overlapping.
Cam: It's this pressure, right, on SaaS to extract value and manage this. So I think the takeaway is these proprietary models are going to be the way this is done, just for cost effectiveness.
Jez: So the question was never whether SaaS survives, it's who ends up with the margin.
And that's your five-ish minutes back. Vibe CFO, Australian founded. An AI CFO for small and medium businesses, out of the box, governed and secure. vibecfo.ai. Now you know.
Now you know.
The AI Exhale is made by VibeCFO. Australian founded, an AI CFO for small and medium business, out of the box.