Jensen Huang: Agentic AI is fully accretive for software companies

Key Concepts

• Project ARK: A new autonomous agent collaboration between NVIDIA and ServiceNow designed to code, reason, and execute tasks within enterprise applications.
• Agentic AI: AI systems capable of autonomous action, tool usage, and reasoning, moving beyond simple generative text.
• The Five-Layer Cake: NVIDIA’s architectural framework for AI, spanning from energy and chips to platforms and applications.
• Identity & Governance for Agents: The necessity of treating AI agents with the same security, access control, and compliance protocols as human employees.
• Real-Time Computation: The shift from static, pre-recorded software to dynamic, real-time processing required by agentic AI.

---

1. Project ARK and Enterprise Automation

ServiceNow and NVIDIA have unveiled Project ARK, an autonomous agent framework. The primary objective is to allow AI to function as an "agentic front door" for enterprises.

• Functionality: The system is designed to code, think, and interact directly with complex enterprise software environments.
• Strategic Goal: To clean up the "messy" nature of enterprise operations by providing a secure, governed environment where AI agents can operate alongside human workers.

2. ServiceNow’s Growth and Financial Outlook

Bill McDermott, CEO of ServiceNow, highlighted the company’s rapid scaling:

• Growth Metrics: Since 2019, the company has grown from $3.5 billion to a projected $16 billion in revenue this year.
• Future Projections: ServiceNow has set a target to reach $30 billion in subscription revenue by 2030, effectively doubling the company’s size again.
• Current Standing: The company reports a $28 billion Remaining Performance Obligation (RPO), which McDermott claims is the fastest growth rate for an enterprise software company at this scale in history.

3. The Evolution of Software and Service

Jensen Huang, CEO of NVIDIA, emphasized that we are witnessing one of the greatest transformations in the history of the software industry.

• Convergence: Huang noted that "Service is software, and software is service." Because the service industry is roughly 100 times larger than the software industry, the integration of AI agents into service workflows represents a massive market expansion.
• The "Agentic" Shift: Unlike traditional software, which is pre-recorded and retrieved from storage, agentic AI must be processed in real-time.
• Computational Demand: Huang stated that the computational requirements for agentic AI have increased by 1,000% compared to generative AI from two years ago. This is due to the AI’s need to read more data, utilize external tools, reason through problems, and generate a higher volume of tokens to perform "useful work."

4. Governance and Security Frameworks

A critical argument presented by both leaders is the necessity of treating AI agents as digital employees.

• Identity Management: ServiceNow aims to manage the identities of both humans and agents. By applying the same "rules and rails" to agents as are applied to humans, enterprises can ensure compliance, governance, and security.
• Regulatory Compliance: As AI agents begin to perform autonomous tasks, they must be provisioned with specific access controls and network permissions, mirroring the infrastructure currently used for human workforce management.

5. Synthesis and Conclusion

The partnership between NVIDIA and ServiceNow represents a shift from "Generative AI" (creating content) to "Agentic AI" (executing work). The core takeaway is that for AI to be viable in the enterprise, it requires a robust infrastructure that handles identity, security, and real-time computation. NVIDIA provides the high-performance compute necessary for the 1,000% increase in processing demands, while ServiceNow provides the operational "rails" to ensure these autonomous agents function within the bounds of corporate governance and compliance. This synergy is positioned as the catalyst for the next phase of enterprise digital transformation.