‘SaaS-pocalypse’ revisited: Three fears answered

STOCKS of Software-as-a-Service (SaaS) companies have rebounded from their February lows, but they are far from unscathed.

As of last Friday’s close of trading, two bellwether SaaS stocks, namely Salesforce and ServiceNow, were still down by more than 25 per cent since the start of the year, and accounting SaaS firm Intuit has fallen by 31 per cent.

The culprit is a narrative dubbed the “SaaSpocalypse” – the belief that agentic AI will render SaaS platforms obsolete.

In February 2026, Anthropic’s launch of Claude Cowork triggered a massive sell-off in the sector. The fear was that autonomous AI agents, capable of handling complex professional tasks ranging from contract reviews to financial analysis, would encroach on the territory of established SaaS platforms.

The panic intensified when Anthropic revealed that it had raised US$30 billion in a new funding round, valuing the company at US$380 billion. The firm’s revenue had skyrocketed from zero to US$14 billion in annualised run rate, growing 10 times for three consecutive years.

It is a compelling story. But there is a difference: While February’s downturn was driven by the fear of the unknown, today, SaaS firms have reported their results and are taking on the market’s fears head-on.

In other words, we have some answers.

Fear #1: Vibe-coding will open floodgates

The first worry is that AI coding agents will make it trivial to build software, flooding the market with cheaper alternatives to established SaaS platforms. The problem with that logic is that technology was never the only barrier to SaaS adoption.

Intuit CEO Sasan Goodarzi laid out a useful framework at the company’s latest earnings call, drawing a line between “context” and “core”.

SEE ALSO

Code is cheap. Trust is not

The SaaS apocalyse: When fear does the thinking

Core refers to Intuit’s proprietary data, domain-specific AI models, and human expertise. Context is where external large language models (LLMs) such as Claude and ChatGPT handle the long tail of industry-specific need, such as a construction company building custom dashboards within Intuit’s enterprise platform.

Here is the key insight: Intuit is not competing with LLM providers. It is partnering with them, and crucially, retaining 100 per cent of the economics.

Veeva Systems, a SaaS firm focused on the life sciences and pharmaceutical industries, tells a similar story. Its CEO Peter Gassner was explicit that LLM providers are infrastructure, not competitors, much as Amazon Web Services (AWS) was to the first generation of cloud software.

Veeva’s AI agents are built using LLMs from Anthropic and Amazon, hosted on Amazon Bedrock. What makes them valuable is not the underlying technology. Instead, it is the compliance-grade, domain-specific intelligence layered on top.

ServiceNow CEO Bill McDermott put it more bluntly: AI is probabilistic, meaning its results are inherently uncertain. Enterprise workflow orchestration is deterministic – predictable and governed.

Said another way, AI does not replace enterprise orchestration. Instead, AI depends on it.

Here is the telling detail: Instead of competing with ServiceNow, Anthropic is partnering with it, helping customers build AI applications on ServiceNow’s platform.

In other words, code is cheap. But trust is not.

Fear #2: SaaS companies will become dumb databases

The second fear is that AI will take over the customer-facing layer, reducing SaaS platforms to basic databases that create, read, update, and delete information. Nothing more.

Intuit’s recent results suggest the opposite is happening. Over three million customers have used its AI agents, with repeat engagement topping 85 per cent.

Why the high repeat usage? Here is a clue: Intuit’s business tax agent is uncovering an average of over US$1,000 in incremental tax deductions per customer. That is real savings in their customers’ hands.

It is no wonder that when AI and human intelligence are offered together, customers’ willingness to pay increases significantly. That is AI making the platform stickier and more valuable. What is more, AI agents are driving higher demand for human assistance, not less.

QuickBooks Live, which offers live support, saw its customer growth exceed 50 per cent in the latest quarter. Furthermore, these users also show a 22-point higher ecosystem attach rate, meaning they consume more of Intuit’s services across the board.

ServiceNow’s advantage illustrates this further. Its configuration management database is essentially a detailed blueprint of how each customer’s company operates – who does what, which systems talk to each other, and how tasks flow from start to finish. These blueprints took years to build and are unique to the needs of each customer.

AI agents cannot figure out how an enterprise works on its own. They need the maps that incumbents have already built. In an AI-powered world, the system of record becomes more important, not less so.

Fear #3: Seat compression will destroy SaaS economics

The third fear cuts closest to the bone: If AI agents handle most of the work, companies will need fewer human users – and therefore, fewer seats to pay for. That is a problem for the SaaS business model, which charges by the number of users.

Salesforce’s latest results provide the most direct rebuttal. The company reported three monetisation paths firing in parallel: Upgrading existing customers to premium AI-powered tiers, expanding seat counts (seven of its top 10 deals included new seat additions), and selling consumption-based flex credits for customer-facing agent use cases.

Meanwhile, its main AI product, Agentforce, has grown from zero to US$800 million in annual recurring revenue in 15 months, with over 29,000 deals closed.

ServiceNow’s numbers back this up. CEO Bill McDermott highlighted an estimated 1.3 billion seats in its addressable market – barely scratched – with its active user base growing 25 per cent year on year in its latest quarter.

Intuit is expanding, too. The accounting software firm’s mid-market solutions grew approximately 40 per cent year on year. Furthermore, the company is looking to grow its direct sales team by around 30 per cent, not exactly the behaviour of a business bracing for fewer customers.

Seat compression, for now, remains a theoretical risk rather than a present reality.

Get smart: Back to basics

Do the answers above mean that the coast is clear? Not so fast. The SaaSpocalypse narrative contains a kernel of truth. AI will still reshape the software industry.

Some SaaS companies will not survive the transition. But here’s what investors should not miss. The way forward requires SaaS companies to return to their roots.

What made SaaS better than legacy licensing models in the first place was a simple but powerful idea: Shifting from a transactional relationship to aligning the company’s revenue to the customer’s needs. You pay for what you use, and the software earns its keep every month.

That alignment has not changed. If anything, AI will test the relationship again.

The SaaS companies that will thrive are those using AI to solve real customer pain points – uncovering tax deductions, accelerating drug approvals, automating IT workflows – rather than trying to shoehorn the technology into areas where it does not belong.

As the data from ServiceNow, Salesforce, Intuit and Veeva shows, the best incumbents are already doing exactly that.

The SaaSpocalypse fear is not over. But real change will likely unfold over years, not weeks.

The writer owns shares of Intuit, Salesforce, ServiceNow and Veeva Systems. He is co-founder of The Smart Investor, a website that aims to help people invest smartly by providing investor education, stock commentary and market coverage