UBS has shifted its stance on the enterprise software landscape, downgrading ServiceNow and effectively removing the last “Buy” rating from its application software coverage. The move signals a growing apprehension among institutional analysts regarding how generative AI will fundamentally reshape the economics of the software-as-a-service (SaaS) industry.
The downgrade, spearheaded by analyst Karl Keirstead, marks a pivotal moment for ServiceNow, a company that had previously stood as the sole outlier in UBS’s software portfolio. While the company remains a leader in workflow automation, the investment bank now views the systemic risks posed by artificial intelligence as too significant to ignore, particularly concerning how software is priced and sold to the enterprise.
For years, the “per-seat” or per-user licensing model has been the bedrock of the software industry. However, the rapid integration of generative AI creates a paradox: while AI makes software more powerful, it similarly makes the human users of that software more efficient. If a single employee can now perform the work of three thanks to AI automation, the corporate demand for multiple licenses may plummet, directly threatening the recurring revenue streams that have driven SaaS valuations to historic highs.
The erosion of the seat-based model
The core of the UBS ServiceNow downgrade AI risks centers on a structural shift in productivity. In a traditional software environment, growth is linear—more employees at a client company mean more licenses sold. Generative AI disrupts this linearity by decoupling productivity from headcount.
Karl Keirstead’s analysis suggests that the industry is entering a period of uncertainty where the “efficiency gain” provided by AI might actually cannibalize the vendor’s revenue. If AI agents can handle routine tasks previously managed by human operators, the total addressable market for “seats” shrinks. This creates a precarious balancing act for companies like ServiceNow, which must find a way to monetize AI features aggressively enough to offset the potential loss of user licenses.
This tension is not unique to one company but is a sector-wide headwind. Analysts are increasingly questioning whether “AI premiums”—the extra fees companies charge for AI-enhanced tiers—can truly replace the steady, predictable growth of seat-based expansion.
Why ServiceNow was the final holdout
Until this downgrade, ServiceNow was the only application software stock UBS maintained as a “Buy.” This preference was rooted in the company’s unique position as a “platform of platforms.” Unlike niche software tools, ServiceNow integrates various business functions—HR, IT, and customer service—into a single workflow engine, making it more resilient to the disruption of any single tool.
The company has been aggressive in incorporating generative AI, launching “Now Assist” to automate everything from incident summarization to code generation. For a time, this agility convinced analysts that ServiceNow could outpace the risks. However, the broader market volatility and the realization that AI might compress the entire software layer have led to a more cautious valuation.
The AI Paradox: Growth vs. Cannibalization
The current conflict in the software sector can be broken down into two competing forces: the “Upsell” and the “Efficiency Gap.”
- The Upsell: Software vendors introduce AI as a premium add-on, increasing the Average Revenue Per User (ARPU).
- The Efficiency Gap: AI allows enterprises to reduce their workforce or stop hiring, leading to a net decrease in the total number of users.
The risk identified by UBS is that the Efficiency Gap may eventually outweigh the Upsell. When the tools become too efficient, the traditional way of charging for them becomes obsolete.
Comparative Shift in SaaS Economics
To understand the shift in how analysts are viewing these companies, it is helpful to compare the traditional SaaS growth engine with the emerging AI-driven reality.
| Metric | Traditional SaaS Model | AI-Driven Model (Risk Era) |
|---|---|---|
| Revenue Driver | User Growth (Seat Count) | Value-Based / Outcome Pricing |
| Productivity Impact | Incremental improvement | Exponential task automation |
| Pricing Stability | High (Predictable recurring) | Low (Transitioning models) |
| Primary Risk | Churn / Competition | Structural License Erosion |
What this means for the broader tech sector
The decision by UBS to depart its application software coverage without a single “Buy” rating is a stark warning to the broader market. It suggests that the “AI hype” phase—where any mention of LLMs boosted a stock price—has ended, replaced by a “show me the money” phase. Investors are now demanding proof that AI can drive revenue without destroying the underlying business model.
This skepticism extends beyond ServiceNow to other giants in the CRM and ERP spaces. If the industry is forced to move toward “outcome-based pricing”—where a company pays for a task completed rather than a user logged in—the transition period will likely be volatile. Such a shift requires a complete overhaul of billing systems, sales incentives, and financial forecasting.
the cost of running these AI models remains high. The compute power required for generative AI puts pressure on gross margins, meaning software companies are facing a double squeeze: potentially lower revenue from fewer seats and higher operational costs to provide the AI services.
Disclaimer: This article is for informational purposes only and does not constitute financial, investment, or legal advice. Investing in equities involves risk.
The next major catalyst for the sector will be the upcoming quarterly earnings reports, where analysts will scrutinize the actual adoption rates of AI premium tiers and any commentary regarding changes to license counts. Market participants will be looking for concrete evidence of “value-based pricing” replacing the aging seat-based model.
Do you think AI will kill the per-user license, or will software companies find a way to charge more for the efficiency they provide? Share your thoughts in the comments below.
