The SaaS Apocalypse That Wasn't: Atlassian's CEO on Software Companies in the AI Era

Will AI Kill SaaS?

As AI rapidly generates code and automates workflows, a pessimistic narrative has spread through the tech world: will traditional SaaS (Software as a Service) companies be completely disrupted by AI? When large language models can spin up an application on demand, who still needs a monthly subscription to Jira, Trello, or Confluence?

This scenario has been dubbed the "SaaSpocalypse." But in a conversation with Atlassian co-founder and CEO Mike Cannon-Brookes, a clear counterargument emerges — this apocalypse hasn't actually arrived.

Atlassian is precisely the kind of infrastructure-layer company that "every other company runs on top of." Its products — Jira, Trello, Confluence — help large teams organize and manage collaboration, build shared organizational knowledge bases, and serve as the underlying foundation of how modern teams work. When the outside world worries that AI will erode the value of these tools, the view from the company's own CEO carries considerable weight.

Why Infrastructure-Layer SaaS Isn't So Easy to Replace

To discuss the fate of SaaS, we need to distinguish between two types of software. The first includes lightweight apps with a single function and simple logic — these genuinely face the risk of being absorbed by AI's generative capabilities. If a well-crafted prompt can replicate the functionality, the justification for a paid subscription starts to erode.

But collaboration platforms like Atlassian belong to a different category. Their value lies not in any single feature, but in years of accumulated organizational data, cross-team workflows, permission structures, and deeply integrated ecosystems. These represent the "memory" and "skeleton" of how a team operates — things that can't be replicated by spinning up an app on the fly.

In other words, the more a software product carries core business operations and the deeper it's embedded in organizational processes, the harder it is to simply replace. AI changes how software is built and how we interact with it — but it doesn't necessarily determine whether software companies survive.

Network effects are a key concept for understanding the resilience of these platforms. When hundreds of employees across an organization are tracking tasks in the same Jira project and consolidating documentation in the same Confluence space, the platform's value grows exponentially with each additional user — every new participant makes the whole system more valuable for everyone else. This effect makes migration costs extremely high: even if an AI tool could generate a functionally similar alternative, it cannot replicate existing historical data, established permission structures, or ingrained team habits. Economists call this phenomenon "lock-in," and together with the data moat it forms a two-layer defense for infrastructure-layer SaaS against disruption. Lightweight SaaS products (like standalone form tools or simple to-do lists) typically lack both of these protections, which is why they're far more vulnerable to being displaced by AI-generated alternatives.

Is AI a Threat or an Amplifier?

From Cannon-Brookes' perspective, AI looks more like an amplifier of capability for companies like Atlassian than an existential threat. Managing large teams and organizing complex information is inherently a hard problem — and AI can be embedded directly into existing platforms to help users retrieve information faster, summarize progress, and automatically generate tasks and documents.

This means platform-based companies have the opportunity to make AI part of their product, rather than being bypassed by it. Having access to vast amounts of structured enterprise collaboration data actually becomes a unique advantage for building intelligent features in the AI age — the data moat hasn't lost its value in the era of large language models. If anything, it's become more critical.

The strategy of treating AI as an embedded product feature rather than an external threat has already produced concrete examples across the industry. Atlassian launched "Atlassian Intelligence" in 2023, integrating generative AI capabilities directly into Jira and Confluence — supporting automatic work summaries, generating JQL queries (Jira Query Language) from natural language, and intelligently surfacing related documents. Similarly, Notion launched Notion AI, Salesforce built Einstein GPT, and Microsoft embedded Copilot throughout the entire Microsoft 365 ecosystem. This trend signals that the dominant response among top SaaS companies is to "internalize AI" rather than compete against it — using their accumulated structured domain data to fine-tune or augment general-purpose LLMs, producing outputs more tailored to specific enterprise scenarios and building a differentiated moat in the process.

Implications for the Industry

The judgment that "the SaaS apocalypse hasn't arrived" has broader relevance for the entire software industry. It's a reminder that narratives about AI disruption are often far too broad. What truly determines the fate of a software company is where it sits in the value stack: does it offer surface-level functionality that can easily be replicated, or is it infrastructure deeply embedded in how an organization operates?

For founders and investors, this also serves as a filtering criterion. When assessing a SaaS company's resilience to AI risk, the right question to ask is whether it controls data, workflows, and network effects that are difficult to migrate away from.

It's worth noting that this article is based on a brief summary of the interview. Many of Cannon-Brookes' specific arguments and data points from the conversation have not been fully disclosed, and the perspectives here primarily reflect his core position.

Conclusion

AI is reshaping how software is built and used — but the prophecy of "AI killing SaaS" is, at least in the eyes of Atlassian's CEO, greatly overstated. Infrastructure-layer collaboration platforms, backed by their data accumulation and deep integrations, are turning AI into an engine for their own growth. The much-debated "SaaS apocalypse" looks less like a total collapse and more like a structural reshuffling of the industry.