ServiceNow: A Risky Bet on AI Agents

This article first appeared on GuruFocus.

It's no secret enterprises operate in complex systems. The average enterprise uses 360 different applications, which leads to siloed systems and disconnected data. ServiceNow (NYSE:NOW) provides a solution to this problem by allowing customers access to all of their data in one centralized platform, enabling customers to build automated workflows to enhance efficiency on top of their existing systems. Because of this layered on top approach, ServiceNow is used by over 85% of the Fortune 500 and nearly 60% of the Global 2000 (10K).

Warning! GuruFocus has detected 3 Warning Sign with NOW. Is NOW fairly valued? Test your thesis with our free DCF calculator.

Due to stellar financials and AI advancements, ServiceNow is currently priced at a premium valuation. However, I believe the current valuation is not justified. It does not account for the risk of mainstream AI agent adoption, the systematic risk of energy demands, a potential price reversion to the mean due to sentiment changes, and overall M&A risk.

The Achilles Heel of AI

ServiceNow's future outlook is directly tied to AI agent adoption. This creates a significant challenge because there is risk with AI agent adoption being too slow and too fast.

The Risk of Being Too Slow

Morningstar and CFRA reports acknowledge a lower-than-anticipated monetization of AI Agents is in the realm of possibilities. The complexity of AI agents could create adoption hurdles for customers since enterprises are bureaucratic. The customers currently using AI products are early adopters, but not all companies are quick to cross the chasm. The current market is partially driven by AI capex spend by hyperscalers, so if the general demand for AI in the midterm does not match expectations, this could cause a bear market if institutions were to take risk off the table all at once.

On the other hand, management mentioned the AI Pro Plus deals (ITSM, CSM, HR) were up over 50% QoQ on the recent earnings call. This shows the strategy to expand beyond IT is working. In 2024, over 50% of the new logos onboarded landed outside of the technology space. This signals that ServiceNow is capable of handling diverse enterprise complexity. (Bank of American Conference Call Transcript).

ServiceNow offers a product called AI Agent Studio that allows customers to create their own custom agentic workflows. This creates a clear path for ServiceNow to boost upsells (Special Call). The company's hybrid subscription and consumption-based pricing model for AI is a key catalyst since this hybrid model creates an opportunity to generate revenue during a contract cycle instead of annual renewals only (JPMorgan Conference Call Transcript).

Story Continues

Despite this momentum, the key insight is that the early success of AI adoption is already baked into the valuation. The risk is not in the present, but in the future since it will be difficult and a lengthy process for mainstream enterprise adoption by the less tech-oriented companies. If momentum slows, the narrative could shift and trigger a sell-off.

The Risk of Being Too Fast

While the risk of slow adoption exists, it's also important to consider the risk of fast adoption. The International Energy Agency (IEA) recently published a report that states, A single ChatGPT query consumes an estimated 2.9 watt-hours of electricity, nearly 10 times the 0.3 watt-hours of a standard Google search.

According to the IEA, it is estimated that inference could account for 80-90% of the total electricity consumption of AI within the next decade. Inference is what occurs when a trained AI model makes predictions or generates outputs. Considering that multiple companies are already competing for limited power grid capacity, this could introduce an uncontrollable constraint for ServiceNow since utility providers are starting to deny and delay requests for new data center connections, citing insufficient capacity on local and regional grids.

Now with that said, ServiceNow is in a differentiated position since it is vertically integrated. It will be exposed to the cost of inference as AI becomes more prominent in enterprise; however, the company operates dozens of data centers globally, giving it a key advantage for supplying demand. Having control and flexibility are valuable since ServiceNow won't be as dependent on third-party data center infrastructure compared to their competition.

While owning infrastructure is a valid differentiator, it does not fully offset the economic reality. Owning the infrastructure directly exposes ServiceNow to the capital and energy costs of the AI. This direct exposure to energy costs could potentially create margin pressure that competitors using third-party cloud infrastructure could avoid.

In short, ServiceNow is in a challenging paradox. If AI agent demand is high, they could face a bottleneck in the near term due to the power grid constraint and experience contractions on multiples and free cash flow. Alternatively, if AI agent demand is less than expected, this could shift the overall sentiment which could impact the valuation and lead to a sharp sell off.

The Current Valuation is Already Priced to PerfectionServiceNow: A Risky Bet on AI Agents

Figure 1: ServiceNow vs. Peers - Enterprise Value / Sales (LTM). Source: Koyfin

There are clear signals showing outperformance is getting hard to sustain. Management recently announced As our business grows, we expect our revenue growth rate to decline over the long term (10K). Following this statement, subscription gross margin fell to 80% in the last reported quarter from 82% a year prior. Full-year guidance for subscription gross margin (83.5%), operating margin (30.5%), and free cash flow margin (32%) remains unchanged indicating management is not expecting significant growth for the remainder of the year as well (Earnings Call Transcript). Considering that institutional ownership of ServiceNow is approximately 91.2%, if sentiment were to turn negative, a rush for the exit by this concentrated ownership base could create selling pressure.ServiceNow: A Risky Bet on AI Agents

Figure 2: ServiceNow vs. Peers - Gross Profit Margin % (LTM). Source: Koyfin

On the other hand, the historical fundamentals support ServiceNow's 115.1 (P/E) ratio. LSEG notes its gross margin and operating cash yield have been higher than the industry average for the past five years reflecting a gross margin (78.5%) and net margin (13.8%) that are significantly higher than the industry averages of 70.2% and 1.3%, respectively. (LSEG Report). ServiceNow also scored in the 98th percentile for financial performance among all Russell 3000 companies, according to ISS-EVA's proprietary model. This ranking explains the current valuation to some extent (ISS-EVA Report).

In January 2025, the board authorized an additional $3B for the share repurchase program, signaling management confidence for the long term as well (10k). ServiceNow has maintained a 98% renewal rate for the fiscal year and has experienced a 12% YoY increase of 2,109 total contracts greater than 1M, from 1,885 the year prior. The valuation is steep for ServiceNow but it is fair to highlight new customers are signing contracts consistently and retention is high.

Excelling beyond the current financials will be extremely difficult to surpass since any miss on expectations could shift the overall sentiment. Lastly, subscription revenue makes up 97% of total revenue. Competitors will likely capture some of this revenue in the long term as AI adoption increases (10k). High margins act as a honeypot for competition, so a reversion to the mean is a reasonable long-term risk to consider.

This Time is Different but Probably Not

ServiceNow recently announced plans to purchase a company called Moveworks for $2.9B, making this their largest acquisition to date. Moveworks is a strategic acquisition because it would allow ServiceNow to offer an enterprise search feature that also enables AI agent automations. For example an employee at a company, could search parental leave policy' in the enterprise search bar. On the backend, ServiceNow would locate the information and then respond with the information requested, but also with an AI agent that could submit the leave request for the employee directly.

While M&A is a well-known risk multiplier, it is worth mentioning 70% of Moveworks customers already use ServiceNow so this reduces this particular acquisition risk (JPMorgan Tech Conference Transcript). ServiceNow has found success with converting M&A deals like this into positive synergies through replatforming, but acquisitions by nature introduce financial strain and risk for the acquirers to some extent. The deal is also undergoing regulatory review by the U.S. Justice Department for potential antitrust concerns so there is a chance it won't close.

Ultimately, while ServiceNow's M&A track record is strong, a multi-billion dollar acquisition is a different class of risk. Integrating such a large acquisition introduces execution risk and financial strain at a time when the valuation demands flawless performance.

Is the Reward Worth the Risk

There is a lot of data that supports ServiceNow's current valuation. Companies with stellar fundamentals, intriguing narratives, and competent management teams tend to result in premium stock prices over time though. Considering that ServiceNow is already priced to perfection, however, I believe the headwinds ServiceNow faces will result in a decreased stock price over time.

If AI agents are not adopted as fast as expected, institutional ownership could decrease, resulting in a sharp price decline. Additionally, the spending required to keep up with competition, hiring, product expansion, and M&A activity will impact the bottom line if any expectations are not met.

ServiceNow is a phenomenal company, and I am bullish on AI agents in the long term. However, from a long-term perspective, I am bearish at this current valuation.

I focus on analyzing foundational companies in the tech infrastructure space. If you found this deep dive on NOW helpful, be sure to follow me for future updates.

View Comments