Which software companies can avoid the AI SaaSpocalypse? | Transcript

Paul Treiber
Hello, and welcome to Strategic Alternatives, the RBC Capital Markets podcast. Please listen to the end of this podcast for important disclaimers. I'm Paul Treiber, Canadian Software Analyst, and I'm joined here today by my colleague Rishi Jaluria, U.S. Software Analyst. We're going to be reflecting on RBC’s Canadian Private Technology Conference, which took place earlier this month. We hosted 25 differentiated private tech companies, and came away with some fascinating insights. We're going to share the headlines with you.

Rishi, it's fair to say that the event was overshadowed by one big topic, which will come as no surprise to anyone who's really following software.

Rishi Jaluria
Yeah, absolutely, Paul. Look, the impacts of AI were reflected in every panel session. What was interesting was actually the lack of consensus. In several cases, there were conflicting views about what the long-term impact on software from AI will be.

Paul Treiber
Yeah, so is software dying? Let's look at the arguments in turn. On the one hand, you have the SaaSpocalypse scenario, you know, one business leader believes AI will self-compound its own development, leading software to converge towards minimal headcount, you know, leaving some software names at risk of going under. On the other side of the argument, others believe that programming employment is growing as models can't replace developers wholesale. Several companies say that their customers are expanding software, not abandoning it. They're using AI to deliver more capabilities and move faster. And also, something that's interesting is that one emerging role for software is optimizing the use of AI tokens, which are extremely expensive, and basically the base unit of text that AI models use to understand language, but in reality the picture is much more nuanced than just the binary death or survival of software.

Rishi Jaluria
Yeah, you're absolutely right, Paul. It seems like businesses with certain models might be better positioned than others. I would call it vertical software firms and regulated workflow, dense environments are better insulated. When we think about data, high complexity, high levels of proprietary data offer the most defence, especially when the data is applied into an intelligence layer, and where it's hard to create, where it has short shelf life, and where it can't be reproduced by off-the-shelf LLMs. A couple examples of this include healthcare EHR platforms, where patient safety demands zero errors, tax research products built on professional content that LLMs can't access, broker-dealer infrastructure requiring years of regulatory navigation. I will also say network effects are also a moat where value lies with the network operator, and it's hard for new entrants to compete. An example of this could be payer-provider healthcare networks, or vendor-to-manufacturer platforms.

Paul Treiber
Another big theme that emerged from the conference was the effect of AI on jobs, and it was interesting that several presenters pointed out that AI is actually not displacing IT budgets, but instead it's cannibalizing labor. There are several software acquirers that expect that their portfolio companies attack customers' labor budgets. One presenter said that tech spend accounts for just $1 for every $20 that's spent on labor. Converting even just a fraction of that into tech could expand the market by multiples. There are numerous examples of this. Several good ones are video security agents replacing security guards, manufacturing software that optimizes tasks to be performed by fewer people.

Rishi Jaluria
Yeah, look, this underlines that it's not just software companies being impacted by AI, but whole industries, and honestly, the global economy. Look, in aerospace, design timelines have collapsed from years of simulation and reviews to days, or even hours. Insurance platforms can compress the entire purchase journey from days of broker callbacks to minutes of digital completion. In fact, many finance participants think financial intermediation is about to collapse, with AI directly connecting supply to demand. At the same time, there is evidence that AI natives are winning against incumbents. For example, legal AI companies are winning enterprise deals by delivering immediate value. A home healthcare software platform is beating legacy EHR platforms.

Paul Treiber
But the progress that AI is making isn't entirely smooth. There are definitely constraints to scaling. Different parts of the AI hardware stack are improving at different rates. For example, processing is growing exponentially, but memory is growing much more slowly. And, in particular, interconnect is a critical bottleneck. In some cases, expensive GPUs are sitting idle because interconnect can't keep pace. Also, power, not capital, is a key constraint.

Rishi Jaluria
And all this disruption continues to hit valuations, and that is having an impact on software M&A. Bid-ask spreads are wide, limiting financial sponsor-to-sponsor transactions. Credit markets for later-stage software deals are muted. The regulatory approval environment is a little bit more favorable, but really any recovery will depend on the recovery of valuations. And look, I'll maybe add that large strategic M&A needs to be more AI-focused than maybe it used to be five years ago, when consolidation could have been a thesis. Now I think it has to accelerate the AI narrative for a large software acquirer.

Paul Treiber
And just finally, there was divided opinion on the rise of sovereign clouds. Now, sovereign clouds are an initiative designed to ensure domestic control over data and cloud computing infrastructure in the public sector and reduce reliance on U.S. cloud providers. Some felt that this would provide a tailwind for Canadian tech companies, with Canada uniquely positioned to meet international appetite for non-U.S. tech providers, however, others feel that open source models will take over in most cases, and that local hosting isn't a meaningful necessity.
Okay, so to conclude there remains a lot of debate around this space and uncertainty of where software will go. Rishi and I have talked at length in the past about moats for software companies, and I think it'd be a great opportunity for Rishi to just run through what he sees as some of the moats here.

Rishi Jaluria
Yeah, look, thanks, Paul. I would say, you know, in addition to what we've been talking about, I think there's really three key themes I wanted to highlight. Number one is really innovation is the biggest moat in software, it's no longer enough to bolt on AI capabilities or have table stakes AI capabilities that can easily be replicated with off-the-shelf LLMs, it really requires a fundamental rethinking and re-architecting, and that's why we continue to believe innovation is the most powerful moat out there. Second is, we believe the market is really under appreciating the value of vertical and domain expertise of understanding the context of the data as it pertains to specific industries, especially mission-critical problems, and being able to solve deep customer problems and workflow problems. We think the market isn't fully appreciating that, and third, I think it's becoming clear that data alone is not as much of a moat as we initially thought. What really matters is how do you take that data and get the correct insights out of it? How do you make it actionable? How do you add context to that and get that deeply embedded within your software solution? I'll turn it back over to you, Paul.

Paul Treiber
Yeah, thanks. Thanks, Rishi, for joining me today on this podcast. To get our full insights on the Top 10 conference themes, contact your RBC representative.

You have been listening to Strategic Alternatives, the RBC Capital Markets podcast. This episode was recorded on June 11, 2026. Listen and subscribe to Strategic Alternatives on Apple Podcasts, Spotify, or wherever you listen to your podcasts. See you all next time.