Software is eating the world, venture capitalist and entrepreneur Marc Andreessen famously declared in 2011. The ensuing 15 years proved him prescient. In February 2026, a Substack article by Citrini Research grabbed headlines and triggered a market sell-off of SaaS (software-as-a-service) firms, wiping out nearly $1 trillion in market value in a matter of days. Citrini’s central thesis? A cannibalistic last feast where AI eats the very software industry that’s been eating the world.

The argument is simple: If anyone can prompt an LLM (large language model) and vibe-code a custom enterprise resource planning or customer relationship management system in an afternoon, then the multibillion-dollar SaaS industry will become a dinosaur overnight. It’s a frighteningly plausible thought that puts the spotlight on Citrini and the article’s author, James van Geelen, but it’s fundamentally naive because it assumes that we live in a world without friction.

Why firms won’t build all systems in-house

AI tools like Anthropic’s Claude are incredible at instantaneous prototyping. But as any software engineer knows, writing code is 10% of the job; maintaining, scaling, and debugging it for edge cases is the other 90%.

Software isn’t just a static pile of logic; it’s a living organism. Production-strength software requires auditability, 24/7 reliability, API (application programming interface) stability, and security compliance, all at once and all at scale. These aren’t things that LLMs, which are random systems by nature, can replicate.

The “AI eats SaaS for lunch” logic naturally leads to the conclusion that all firms will build all their software in-house—because now, with the help of AI, they can. But can they really? Will they really? I bet they won't for two reasons.

For starters, a primary reason why companies buy enterprise software is to transfer risk. When a Fortune 500 company uses specialized software by a SaaS provider for cybersecurity or HR, they aren’t just buying code; they’re buying compliance with security frameworks, GDPR (general data protection regulation) indemnity, and ISO (International Organization for Standardization) certifications. If firms build systems in-house using general-purpose tools such as Claude or Google’s Gemini, what happens if (or when) things go wrong (such as data leaks)? There will be no vendor to sue, no platform to blame, and no security patch to purchase.

Another reason relates to scale and interoperability. A third-party provider spreads the cost of high-level security and compliance across thousands of customers. An individual firm trying to replicate that in-house would find the “efficiency” of AI quickly eaten up by the massive overheads of self-certification and liability insurance.

If all firms build all their systems in-house, we’ll be back in the world of fragmented software with limited interoperability. Remember legacy systems? Firms will end up with isolated legacy piles of AI-written code that no one understands.

Opportunities and challenges

If the trend of building software in-house actually takes off, and every company starts creating its own bespoke AI systems, the complexity of auditing those systems becomes exponential. Auditors would then become the most needed and sought-after professionals on earth. If van Geelen really believes what he says, then he might consider auditing as a new profession to hedge against the apocalypse he hypothesizes. Who knows, the auditing profession might be the answer to the problem of AI displacing jobs.

Summarizing all the above, let’s not forget that AI is a general-purpose technology, and software companies are specialists. It’s hard to argue that generalists will replace all specialists in the modern world, which is essentially what Citrini’s scenario argues.

Everyone can buy a shovel. Not everyone shovels their own snow.

What’s more, the “AI eats everything” narrative assumes that LLMs have access to all the world’s intelligence. They don’t. Besides every company’s proprietary data, there’s the paywall barrier, which keeps the most valuable data—highly structured, organized and predictable information required for professional-grade decisions—behind the moats of companies like LexisNexis, Thomson Reuters, Nielsen, and the like.

Without access to this information, generic models can’t generate deep insights; they risk simply recycling data that’s available in the public domain. Indeed, the owners of the data, not the owners of the models, hold the ultimate leverage.

What will AI do to SaaS?

SaaS won’t be eaten by AI, but will be shaken and stirred by it. For instance, Block reduced its head count by nearly half in February, culling more than 4,000 positions. And in March, Atlassian, one of the SaaS companies hit hardest by the market sell-off, retrenched 10% of its staff.

Gone is the cozy convention of per-user pricing and 5% annual price increases justified with the release of new features nobody asked for. Companies will need to be more discerning and can wield a credible threat to take business away from an SaaS provider. Whether they follow through on it is another matter.

This threat will inevitably shake up the SaaS industry. Those who can deliver measurable value will survive and thrive. Those who can’t will perish, and the industry will emerge leaner and stronger.

A version of this article was published in The Business Times. This article was published April 27, 2026, by INSEAD.