Agentic AI: Beyond Nvidia in the Investment Cycle

A new chapter is beginning in the AI investment cycle, shifting focus beyond infrastructure to agentic AI. These are AI systems that can perform complex, multi-step tasks autonomously, without constant human direction. They can plan, reason, take actions in software, and adapt to new information. This means they are more than just chatbots; they can browse the internet, execute code, and manage workflows. While much of this is still aspirational, the direction is clear, and investment implications are emerging. The infrastructure layer, like NVIDIA's GPU clusters, has seen massive growth. NVIDIA alone captures about 90% of AI accelerator spending, which is roughly $180 billion annually. Data center and networking companies have also seen demand surge, with growth in data center equipment expected to continue at 25% annually for the next four to five years. The big question now is whether the application layer can generate enough revenue to justify this infrastructure investment. The focus is shifting to enterprise software, including AI-native applications and agentic systems that reduce labor costs or increase revenue. Early evidence of this enterprise monetization is appearing, particularly in cybersecurity, with AI-powered threat detection and automated incident response moving into production. This shift means new opportunities are emerging for companies developing these AI applications.