Framework Fortresses: How Infrastructure Secures Its Place in the AI Agent Stack

Key Insights

• Autonomous AI agents, now capable of iterative reasoning, tool usage, and sophisticated decision-making, are emerging as practical, data-driven problem-solvers across different verticals.
• While individual agent tokens may capture fleeting attention, their long-term value is more uncertain; the foundational infrastructure that supports many agents simultaneously will likely maintain more durable growth.
• ai16z and Virtuals demonstrate how the infrastructure layer can build lasting competitive advantages through network effects, developer engagement, and value accrual—ultimately creating moats that standalone agents struggle to match.
• As the space quickly evolves, individual agents may morph into frameworks with take rates themselves.

The size of the SaaS market is estimated to be around $330 billion and expected to grow to $1 trillion by 2032. Y Combinator says that the vertical AI agent market could be ten times bigger. The rise of autonomous AI agents represents one of the most significant shifts in the digital economy. These agents—capable of executing complex tasks, reasoning iteratively, and leveraging APIs—are quickly becoming important players in areas ranging from creative industries to consumer-facing applications. Enabled by improving foundational models and agentic workflows, agents are no longer just simple chatbots. They can now plan, improve upon their own outputs, and interact with ecosystems of other agents, protocols, and data sources. As a result, they are quickly moving from amusements into powerful, specialized problem-solvers. Agents will conduct onchain transactions for us through natural language prompts (like Griffain), act as a seamless interface between us and varying applications (like Orbit), and become tireless research assistants who can sift through and distill incoming information (like aixbt).

However, as crypto investors consider where best to place their capital in this shifting landscape, it is essential to distinguish between excitement surrounding individual “agent tokens” and the more foundational layers of the agent stack. While singular agents may achieve brief cultural relevance, their long-term value is less certain. In contrast, the underlying infrastructure—encompassing open-source frameworks, platform protocols, and the orchestration layers that enable agent interaction and scalability—holds more durable growth potential.