Industry Analysis8 min read
The Great SaaS Unwinding: Why AI Agents Are Breaking the Per-Seat Pricing Model
For a decade, Software-as-a-Service (SaaS) enjoyed Wall Street's most lucrative business model via "per-seat" subscriptions. But as AI agents take over software operations and compress human headcount, SaaS companies are facing a severe revenue contraction and a reckoning in their market multiples.
The Headcount Paradox and Software's Fundamental Conflict
For more than a decade, the Software-as-a-Service (SaaS) industry has relied on a simple, incredibly powerful flywheel: convince enterprises to license software and charge them on a "per-user, per-month" basis. This model worked flawlessly during an era of labor expansion, generating highly predictable revenue and driving ARR (Annual Recurring Revenue) multiples upward of 15x.
However, according to real-time data analysis across Tech and TIKHUB, 2025 has marked a fundamental tipping point. The maturation of AI agents means software is no longer just "assisting human work" but "executing human work." This introduces a fatal conflict of interest: the more powerful your software becomes, the fewer human employees your customer needs. And if your customer needs fewer employees, your "per-seat" revenue collapses.
Recent reports indicate that over 85% of forward-looking software contracts are now shifting toward usage-based or outcome-based hybrid pricing models. Incumbent SaaS giants stubbornly clinging to the old per-seat orthodoxy are being severely punished by the market, seeing their multiples compressed to 6x ARR or lower.
Charlie Munger’s Inversion: When Peak Efficiency Means Zero Revenue
Legendary investor Charlie Munger often advised: "Invert, always invert." If we apply this mental model to the current SaaS landscape, a massive structural risk is exposed.
Suppose the mission of a B2B SaaS company is to "maximize the productivity of the client's team." In the past, this meant providing a better UI so 100 customer support reps could work faster. Today, an AI agent can allow 10 human reps to accomplish the work of those 100. For the customer, this is a massive efficiency gain. But for the SaaS company charging per seat, 90% of their licensing revenue vanishes overnight.
Munger’s wisdom reminds us that any business model where the realization of customer value (reducing headcount) directly destroys the company's growth engine (adding seat licenses) is ultimately unsustainable. This fundamental misalignment is exactly why many "AI-enabled SaaS" plays are turning into value traps.
Howard Marks and the Multiple Compression Pendulum
Oaktree Capital's Howard Marks frequently uses the "pendulum" analogy to describe markets. In 2020-2021, the SaaS pendulum swung to an extreme of optimism, pricing in infinite growth. Today, faced with higher interest rates and the deflationary nature of AI, the pendulum is swinging back violently.
The key question for investors is distinguishing whether this is merely a cyclical tightening of enterprise IT budgets, or a structural paradigm shift. Marks emphasizes evaluating the underlying cash-flow-generative power of the asset. If traditional SaaS companies cannot rapidly pivot to billing for API calls or compute resources, their future free cash flow will face permanent impairment.
Investors must tread carefully. On the surface, a software stock that has fallen from 15x to 5x ARR might appear to offer a deep margin of safety. But if its billing foundation is being hollowed out by AI agents, that 5x multiple could still be too expensive.
Warren Buffett's Moat Test: The Commoditization of the UI
Warren Buffett defines a great business by its "economic moat." For enterprise software, this moat was historically built on "switching costs." Employees formed habits around a specific user interface (UI), making the retraining of an entire workforce too costly for a client to switch vendors.
AI agents are changing the rules of the game. The future user of enterprise software is not a human, but an AI agent communicating via APIs and working in the background. When the "operator" becomes a machine, the competitive advantage of a sticky UI/UX disappears. AI does not care about beautiful dashboards; it only cares about API latency and data accuracy.
This implies that software services risk degrading from a high-retention "application layer" into a commoditized "infrastructure layer." Once switching costs drop, the SaaS company loses its pricing power. And as Buffett warns, a business without pricing power cannot sustain excess returns over the long term.
Investor Strategy: Seeking the New <a href="/glossary/intrinsic-value">Intrinsic Value</a>
How should value investors adjust their frameworks in the face of this great unwinding?
First, when building a Discounted Cash Flow (DCF) model, higher terminal decline rates must be applied to companies reliant solely on per-seat billing. Second, investors should seek out businesses that have successfully transitioned to "hybrid" or "consumption-based" models. If a software company can charge based on "tasks completed" or "compute consumed," the explosion of automation driven by AI agents becomes a massive tailwind rather than a threat.
The AI productivity revolution is real, but not all of the legacy winners will capture its value. Companies that possess true pricing power and the agility to transform their business models will be the economic castles of the next decade.
Recent posts
%22%2F%3E%3Cpath%20d%3D%22M0%20610%20370%20370l300%20160%20400-330%20530%20360%22%20fill%3D%22none%22%20stroke%3D%22%23c99a62%22%20stroke-width%3D%2218%22%20opacity%3D%22.65%22%2F%3E%3Ccircle%20cx%3D%221070%22%20cy%3D%22200%22%20r%3D%2246%22%20fill%3D%22%23c99a62%22%2F%3E%3C%2Fsvg%3E)
%20-1.jpg?width=1600)
Frequently asked questions
What is Per-Seat Pricing?
It is the most common billing model in the SaaS industry, where a customer pays a fixed monthly or annual subscription fee for each human employee using the software.
Why do AI agents threaten this model?
AI agents can automate vast amounts of work previously done by humans. As companies require fewer human employees to operate software, they need fewer "seats," directly shrinking the recurring revenue of the SaaS provider.
What is a Hybrid or Consumption-based model?
It is a pricing structure that combines a base subscription with variable billing tied to actual system usage, such as API calls, compute resources, or tasks completed. This allows software companies to monetize the high-frequency actions of AI agents.
Stop trading on demand headlines — stress-test your SaaS holdings with master investing frameworks and DCF models in Valuefocus:
Download Valuefocus