Stories
ServiceNow Executive Amit Zavery Says Frontier AI Companies Lack a Durable Moat
- •Amit Zavery of ServiceNow argues that markets are misinterpreting software by lumping all companies together amid the AI boom.
- •Frontier AI models are compared to interchangeable infrastructure like chips, with customers prioritizing outcomes over specific underlying models.
- •ServiceNow's competitive advantage stems from deep enterprise context, workflows, and governance rather than the LLM itself.
- •Maintaining enterprise software accounts for 80% of costs, making domain expertise and security crucial beyond just writing code.
- •Pricing models are shifting from traditional seat-based licenses to outcome and consumption-based models.
Key Questions Answered
Amit Zavery argues that frontier AI companies do not have a durable moat because large language models resemble commoditized infrastructure like chips. ServiceNow can flexibly switch between models based on cost and performance, as customers primarily care about outcomes rather than the specific underlying model.
According to Amit Zavery, less than 10% of the intellectual property inside ServiceNow's products comes from the LLM. The remaining majority stems from years of deep operational context, workflow understanding, business requirements, and governance.
Rather than eliminating jobs through automation, ServiceNow has shifted employees into newer roles such as AI operations, security, analytics, and managing multiple AI agents. Employees who previously worked on the IT help desk were retrained and redeployed to oversee AI systems, handle complex escalations, and refine how those agents operate.
