Parasail Raises $32M to Build an AI Supercloud for Agent Deployment

SAN FRANCISCO: Parasail has raised $32 million in a Series A round to scale its AI inference and training platform for agent deployment. The round was co-led by Touring Capital and Kindred Ventures. Samsung NEXT, Flume Ventures, and Banyan Ventures also participated. Total funding reaches $42 million.

Parasail launched in April 2025 and now processes over 500 billion tokens per day. It describes itself as an AI Supercloud, an orchestration layer that aggregates GPU capacity across 40 data centres in more than 15 countries and automatically optimizes endpoints for speed, cost, and performance. Developers can deploy custom AI models in under five minutes. No long-term contracts required.

The pay-per-token model is the core differentiator. Enterprise GPU contracts typically require multi-year procurement commitments that most startups and early-stage AI teams cannot absorb. Parasail removes that barrier, buying compute on demand across a global supply network and routing workloads automatically.

The timing is deliberate. AI agents chain multiple models together, generate large and unpredictable token volumes, and require infrastructure that can scale instantly without manual configuration. Parasail targets that runtime layer directly. Customers include Elicit, mem0, Gravity, Kotoba, and Venice. Monthly revenue is growing 30% month-on-month.

CEO Mike Henry previously built Groq’s cloud offering. The $32 million will deepen inference and training orchestration, accelerate go-to-market, and strengthen partnerships across the GPU and data centre ecosystem.