Introduction

The AI agent ecosystem, valued at $17.4 billion and encompassing over 1,250 agents, is dominated by four core verticals: frameworks, launchpads, AI agents, and swarms. Frameworks, like AI16Z’s ElizaOS, serve as foundational SDKs for developing AI agents, akin to Ethereum’s EVM for Layer 1s. Launchpads, such as Virtual on Base and Vvaifu on Solana, simplify agent deployment, mirroring the role of rollup-as-a-service providers in crypto. AI agents, the most consumer-facing vertical, are used for diverse applications like DeFi, social media management, and gaming, leveraging tools like ElizaOS and Virtual. Emerging swarms, composed of interoperable agents, represent a collaborative evolution, enabling agents to tackle complex tasks collectively. This layered ecosystem showcases significant potential, as frameworks, launchpads, and agents increasingly blur lines and integrate.

While Solana and Base currently dominate the market with a combined 98% share, their stronghold is poised for disruption. Solana leads in AI agents with a 67% market share and a more distributed network of mid-market players, while Base excels in infrastructure with a concentrated market cap reliant on top projects like Virtual and Cookie DAO. New entrants like Hyperliquid and Sui are expected to challenge this duopoly, potentially reshaping the landscape. As AI agents and infrastructure evolve, emerging chains will likely adopt Solana’s dispersed or Base’s concentrated approach, influencing how market share and technological innovation unfold. This dynamic ecosystem is set to see increasing cross-vertical integration, driving collaboration and growth across chains.

The competitive landscape dominated by Solana and Base is shaping the trajectory of the broader AI agent ecosystem. As the market evolves, these chains' differing approaches—distributed versus concentrated—will likely influence the development and adoption of key components across the ecosystem. This transition opens the door to understanding the foundational layers of the AI agent ecosystem, including frameworks, launchpads, AI agents, and swarms, which collectively form the building blocks of innovation and integration in this space.

The AI Agent Ecosystem

The AI agent ecosystem can be divided into four primary layers:

Frameworks
Launchpads
AI Agents
Swarms

1. Frameworks

Frameworks are like SDKs tailored for AI agents, forming the foundational layer of the AI stack. They empower developers to design, deploy, and manage autonomous agents efficiently. Think of them as analogous to Ethereum’s EVM for launching Layer 1s or Optimism’s OP Stack for Layer 2s.

This is one of the most prominent AI agent verticals. The leading framework is AI16Z’s ElizaOS, a TypeScript-based multi-agent simulation framework. It gained significant traction last December, becoming GitHub’s #1 trending repository with over 4,000 stars and 1,000 forks.

Other notable frameworks include:

G.A.M.E.: A low-code toolkit developed by the Virtual team.
Zerepy: An open-source Python-based framework by the Zerebro team.
Dolion, ARC Rig, and Aether: Emerging players in the space.

2. Launchpads

Launchpads have quickly become the most popular vertical in the Crypto x AI intersection, spurred by the explosive rise of Virtual last year. At their core, launchpads are platforms designed to simplify the creation and deployment of AI agents. They function similarly to Rollup-as-a-Service (RaaS) providers like Conduit or Pumpfun for memecoins.

In fact, Virtual brands itself as “Pumpfun, but for AI agents.” Given Pumpfun’s success in the memecoin niche, this parallel is compelling. Virtual stands out as the most successful launchpad and AI agent to date. It’s a no-code AI launchpad built on Base and also owns the G.A.M.E. framework.

Other notable launchpads include:

Vvaifu on Solana
Vapor on Hyperliquid
Daosdotworld, Creatorbid, and Holoworld

These layers collectively form the backbone of the rapidly evolving AI agent ecosystem.

3. AI Agents

This is the most consumer-oriented and competitive segment within the Crypto x AI vertical. AI agents are deployed via launchpads like Virtual or built using frameworks like ElizaOS. For example:

AIXBT: An agent launched through Virtuals.
DegenSpartanAI: Built using ElizaOS.

These agents perform highly specialized tasks such as:

Automating investment strategies (e.g., AI16z).
Managing social media accounts (e.g., LUNA).
Providing unique virtual personalities for productivity or entertainment (e.g., ZEREBRO).

This is the broadest stack in the AI agent ecosystem and can be further divided into subcategories:

DeFi and Trading Agents: Griffain, Hey Anon, etc.
KOL Agents: Dolos, AIXBT, LUNA, etc.
Investment DAOs: Vader, AI16Z, AImonica, Sekoai, etc.
Music & Art: ZEREBRO, nft_xbt, Sandy, etc.
Gaming & Metaverse: Hyperfy, Illuvium, ARC, etc.
Developer Utilities: Soleng, Certaik, etc.

Cross-Vertical Integration:

The lines between these verticals often blur:

Some agents own launchpads (e.g., Virtual’s G.A.M.E, Dolos’s Dolion).
Others evolve into frameworks (e.g., ZEREBRO’s Zerepy, ARC’s ARC Rig).
This dynamic allows agents to extend their capabilities, creating a loop where frameworks, launchpads, and agents evolve together.

4. Swarms

AI Swarms represent the next evolutionary step, enhancing the capabilities of individual agents through collaboration. These swarms, made up of interoperable agents, tackle complex tasks that single agents cannot handle efficiently. Whether frameworks evolve into launchpads or vice versa is secondary; what matters is the emergence of swarms. Swarms enable agents built on specific frameworks or launched via particular launchpads to:

Communicate,
Coordinate, and
Collaborate on tasks.

For example:

Vertical Integration Swarms: Teams owning frameworks and launchpads, such as Virtuals or Vvaifu, facilitate agent collaboration within their stacks.
Interoperability Layers for Swarms: Technologies like FXN, Swarms, and Theoriq enable cross-framework or cross-launchpad swarms.

Although less popular than other tech stacks, swarms are often bundled into frameworks. Think of swarms as interoperability layers for agents and frameworks.

Other Crypto x AI Verticals

Beyond the main stacks, the ecosystem also includes:

Data Providers: Cookie DAO, Sentient Markets.
Compute Providers: Akash, Render.
Model Creation: Bittensor, Nous Research.
Verification & Privacy: Phala Network, Nillion.

Suiai is a Virtuals-like launchpad for other blockchains, while NODO’s project spans the AI agent and launchpad categories with plans for future expansion.

AI Agents

The entire AI agents category is valued at $17.4 billion, tracking over 1,250 agents, as per Cookie DAO. Solana dominates with a 56% market share, amounting to a market cap of $9.83 billion, while Base follows with a 40% market share and a $7 billion market cap. Together, Solana and Base account for 98% of the AI agent market landscape.

If we examine the top 10 agents by market share, Solana claims seven spots, Base two, and Ethereum one. When grouped by market cap, Solana’s top 10 agents total $4.3 billion, while Base’s top 10 come to $4.47 billion. This indicates that while Solana has more agents, Base’s agents are more concentrated in value.

Base’s overall market cap for AI agents is $7 billion, with the top 10 agents accounting for 64% of its market share, equivalent to $4.47 billion. In contrast, Solana’s total market cap of $9.38 billion sees its top 10 agents contribute only 44% of the share. Base’s fewer, high-value agents contrast with Solana’s broader pool of mid-market cap agents.

The dominance of Solana and Base, holding 98% of the market share, is expected to decline as more chains enter the AI agents space. Likely challengers include Hyperliquid and Sui, emerging as the next prominent chains after Solana and Base. Outside AI agents, Solana continues to lead as the hottest Layer 1, and Base holds the title for the hottest Layer 2 over the past 12 months. By "hot," this refers to chains where users are actively engaging.

Base’s market cap share among agents is expected to become less concentrated over time, resembling Solana’s. As new launches and smaller agents gain traction, Base’s top 10 share may decline from 64% to around 50%.

Agents

From our categorization, agents and infra represent the two overarching groups. Frameworks and launchpads are categorized under infrastructure, while AI agents and swarms fall under agents.

In terms of agents, they dominate the overall market cap with $9.29 billion, representing 53% market share. Of this, 67% resides on Solana ($6.24 billion) and 31% on Base ($2.87 billion). Solana leads Base significantly in terms of agent count and market share.

Among the top 10 agents, Solana claims six spots, Base three. Across all chains, the top 10 agents hold 40% of the total market share, translating to $3.73 billion in market cap. This indicates a less concentrated market, with a broad distribution of mid-market cap agents.

This dynamic is partly due to the relative ease of launching agents. However, as time progresses, market concentration is likely to increase, driven by power laws. Many agents may fail to meet expectations and eventually go to zero, consolidating the market around a few winners.

Note: This ranking focuses on market share, not market cap.

Infra

In our framework, infrastructure (infra) includes categories such as frameworks and launchpads. This vertical holds a $7.38 billion market cap, representing 42% of the AI agents category. While agents lead in market cap, share, and count, launching a successful infra project is relatively harder than launching an AI agent or swarm.

Although Solana leads in agents, Base takes the lead in infrastructure, largely driven by Virtuals, which accounts for 87% of its infra-market cap. When combined with Cookie DAO, these two projects represent 91% of Base’s infra-market cap. In contrast, Solana’s top eight infrastructure products account for 79% of its market cap share, indicating greater diversification.

Virtuals and Cookie DAO are the only Base projects in the infra top 10, while Solana holds eight spots. Virtuals’ 87% dominance in Base’s infra category is likely to diminish over time. The 91% share currently held by Virtuals and Cookie DAO may eventually be spread across 5–7 projects as the market diversifies.

Key takeaways

Solana and Base's dominance in market share, currently at a combined 98%, is likely to face disruption as AI agents emerge on other chains like Hyperliquid and Sui. This substantial market share offers significant opportunities for these upcoming chains to capture a portion of the total addressable market.
Solana leads in AI agents, leaving Base room to expand in this category. Conversely, Base dominates in infrastructure, providing Solana an opportunity for growth in that area.
Base’s AI agents and infrastructure are concentrated among a few key players, whereas Solana’s market presence is spread across a broader range of projects. However, Base’s concentration is likely to be challenged as more successful AI agents emerge on the chain.
New chains might adopt strategies similar to Base or Solana—either focusing on a few large winners like Base or diversifying with numerous mid-sized successes like Solana. It will be fascinating to see how Hyperliquid and Sui approach this landscape and whether they replicate or redefine these playbooks.

Closing Thoughts

In conclusion, the AI agent ecosystem is at a pivotal moment, with Solana and Base leading the charge but facing imminent disruption from emerging players like Hyperliquid and Sui. As the ecosystem matures, its layered structure of frameworks, launchpads, AI agents, and swarms will continue to drive innovation and integration. Whether through Solana’s diversified approach or Base’s concentrated model, the strategies adopted by new entrants will determine how market share evolves and value consolidates. With the lines between infrastructure and agents increasingly blurring, the ecosystem is poised for a transformative phase, marked by heightened competition, collaboration, and technological breakthroughs. The next era of AI agents promises not only to reshape the competitive landscape but also to unlock new frontiers of possibility across chains.