The Twilight of SaaS: Native AI Verticals Take Over

What Happened?

January 27, 2025, will go down in history as the day the classic SaaS model received its death certificate. That Monday, the SaaS market lost $300 billion in a single session, a crash that, according to Richard de Silva, founder of Lateral Investment Management, marks the end of the classic SaaS model (Crunchbase News, 2025). The immediate cause was the emergence of AI agents that act without human intervention — the so-called 'headless' model — which dismantles the fundamental premise of per-seat pricing. Companies like Salesforce and Workday, which built empires on that scheme, see their revenue predictability evaporate.

To understand the magnitude of the event, it's worth remembering that the SaaS market had been a haven for investors for two decades due to its recurring, scalable, and predictable subscription model. Salesforce, founded in 1999, was a pioneer in demonstrating that cloud software could generate stable and growing revenue, reaching a market capitalization of over $200 billion at its peak. Workday, meanwhile, revolutionized enterprise management software with a similar model. But the arrival of generative AI, and particularly autonomous agents capable of performing complex tasks without human supervision, has broken that paradigm. If an AI agent can draft contracts, analyze balance sheets, or manage inventories, why pay for each human user? The market has responded with a massive sell-off of traditional SaaS stocks.

Why It Matters

The change is not incremental: it is structural. Generic, horizontal SaaS is becoming a declining legacy, as on-premise software did before. In its place, native AI vertical platforms are emerging, designed for specific industries (legal, finance, healthcare) and charging for work done or results. For example, an AI legal assistant bills per contract drafted, not per license; expense management software takes a percentage of the savings detected. This expands the addressable market from IT budgets to the much larger labor budgets, estimated at $2 trillion in white-collar services alone (Crunchbase News).

This model shift has profound implications. In traditional SaaS, growth was measured by the number of seats sold; in the new paradigm, value is measured in time saved, errors avoided, or revenue generated. Companies that can demonstrate a clear return on investment will capture a significant portion of that $2 trillion market. Moreover, verticalization allows for proprietary data moats: an AI platform for medical diagnostics, for example, becomes more valuable the more cases it processes, creating a positive feedback loop that horizontal competitors find hard to replicate.

Historically, we have seen similar transitions. The shift from on-premise software to cloud SaaS eliminated giants like Siebel Systems and PeopleSoft, which failed to adapt. The difference now is speed: while the transition to SaaS took over a decade, AI is compressing cycles into months. According to PitchBook data, funding for vertical AI startups quadrupled between 2023 and 2024, reaching $45 billion, while traditional SaaS rounds fell 30% in the same period.

Consequences for Companies and Users

• For traditional SaaS companies: They must urgently pivot to usage-based or outcome-based pricing models, or they will disappear. The adaptation window is narrow. Salesforce has already announced a new product, Agentforce, which charges per completed action, not per user. However, its legacy customer base, accustomed to per-seat pricing, may resist. Workday, meanwhile, has launched an AI platform for HR that automates recruitment processes, but has not yet detailed its pricing model. Regulatory uncertainty also plays a role: in the EU, the AI Act could require algorithm transparency, increasing compliance costs for these companies.
• For customers: Software costs will be directly tied to value generated, which can reduce fixed expenses but introduces budget uncertainty. A company hiring an AI legal assistant will pay only for contracts drafted, but if workload fluctuates, so will spending. This may benefit SMEs with variable workloads, but large corporations, which prefer predictable costs, will need to adjust their financial planning models. Additionally, reliance on proprietary data poses vendor lock-in risks: once a company trains a model on its data, migrating to another provider can be costly and complex.
• For investors: The new wave of vertical AI startups offers potentially higher returns, but with different risk profiles, depending on proprietary data and niche markets. A venture capital fund investing in a dozen vertical AI startups can expect some to become unicorns, but others will fail if they cannot scale beyond their initial niche. The valuation of these companies is no longer based on metrics like ARR (Annual Recurring Revenue) or churn rate, but on the depth of their data moat and the efficiency of their agents. This demands a new way of evaluating investments, something many traditional funds have yet to master.

What You Need to Know

SaaS is not dead, but its dominant form is. The next decade will see software fragment into hundreds of vertical AI solutions, each with its own data moat. Companies that invest in proprietary data and specific process automation will lead. For professionals, the critical skill will be understanding how to integrate these tools without losing strategic control. As de Silva notes, 'companies that adapt will build faster and deliver more value.'

A concrete example: in the legal sector, startups like Harvey (backed by OpenAI) are already replacing junior legal assistants in document review tasks. Harvey charges per query, not per license, and has reported a 40% reduction in contract review time. In finance, the platform Brex uses AI to automate expense management, charging a percentage of the savings detected. These companies are growing at rates exceeding 200% annually, while traditional SaaS stocks fall.

For IT professionals, the recommendation is clear: start evaluating which processes in your organization can be automated by AI agents, and seek vertical providers specialized in your industry. Integration will require skills in APIs, data security, and AI governance. For investors, the time to rebalance portfolios is now: reduce exposure to horizontal SaaS and increase in vertical AI startups with proprietary data.

«The death of per-seat pricing is the first symptom of a transformation that will redefine the entire software industry.» — Richard de Silva, Lateral Investment Management

In summary, January 27, 2025, was not a day of irrational panic, but an early warning signal. Classic SaaS, with its per-seat pricing model, is being replaced by an ecosystem of AI agents that charge for results. Those who do not adapt will be left behind. History has taught us that technological transitions are relentless: Blockbuster did not survive Netflix, and Nokia could not compete with the iPhone. Now, traditional SaaS faces its own Blockbuster moment. The question is: which companies will be the Netflix of this new era?