Parasail raises $32M Series A to build a supercloud that lets developers control AI

Touring Capital and Kindred Ventures co-lead $32 million round to expand Parasail’s AI supercloud, an inference and training platform built for AI agents

Parasail, the company building the AI ​​Supercloud, an inference and training platform for deploying and scaling AI agents, announced it has raised $32 million in Series A funding, bringing total funding to $42 million. The funding round was co-led by Touring Capital and Kindred Ventures, with participation from Samsung NEXT, Flume Ventures, Banyan Ventures, and existing investors.

The new funding will be used to expand Parasail’s AI Supercloud, a fabric of global computing resources that automatically optimizes model endpoints for speed, performance, and cost. With the additional funding, Parasail will deepen its orchestration and inference optimizations, accelerate go-to-market efforts, and strengthen strategic partnerships across the GPU and data center ecosystem.

The world is currently rebuilding the entire cloud around AI, spending trillions of dollars to build data centers and equip them with GPUs. However, developers are still constrained by access to this infrastructure and the challenge of getting AI models up and running for products quickly and efficiently. From vertically centralized enterprise agents to new consumer-centric personal agents to a broad agent SDK platform, Parasail powers inference and reinforcement learning environments for the rising wave of AI agents that disrupt traditional application paradigms.

Parasail provides an AI supercloud that meets the insatiable need for customized, instant, reliable inference and continuous training. Customers can now set up highly scalable AI in less than 5 minutes, leveraging the world’s supply of GPU computing.

AI as a component of modern software

The developer-controlled AI market is rapidly expanding, with analysts estimating it to exceed $100 billion as startups and enterprises move toward aggressive performance goals, custom models, and distributed computing ecosystems. A vast software, hardware, and data center ecosystem is forming to support that change.

Companies are realizing that AI has become a component of modern software they can deploy, leading to an explosion of open source and specialized models. To unlock the potential of AI, software companies need greater control over cost, latency, and customization, so they are moving away from black-box APIs and toward self-operated AI endpoints. But the infrastructure behind that change remains fragmented. GPU supply is constrained and inconsistent, inference optimization is complex, and scaling often requires contracts, sales negotiations, and months of integration effort.

Parasail removes that friction and provides a seamless customer experience for AI agent deployment with production-ready AI endpoints in minutes and scales to large-scale traffic without contracts or infrastructure management. With five lines of code, developers can launch custom models to meet aggressive latency, throughput, and tokens per second goals, and easily handle large spikes in traffic when their product takes off.

“AI builders don’t need to be infrastructure experts to ship great products,” said Mike Henry, founder and CEO of Parasail. “AI is becoming the core infrastructure of modern software, but the infrastructure layer itself hasn’t kept up. We built Parasail to enable teams to deploy custom AI at scale without negotiating contracts, managing fragmented GPU supplies, or hiring performance engineering teams.”

Also read: AiThority Interview with Glenn Jocher, Ultralytics Founder and CEO

Structural layers of AI inference

Parasail is not just another single cloud provider. Operate a programmable deployment network that abstracts supply fragmentation and inference optimization.

The differentiation is structural, and the model allows startups and growth-stage companies to scale from zero to enterprise-grade workloads without having to rewrite their infrastructure as demand increases.

• Developer-first deployment: Production AI endpoints are up and running in minutes with minimal code, abstracting significant backend orchestration complexity.
• Economics as a first principle: Rather than competing on discrete speed benchmarks, Parasail optimizes workloads for cost efficiencies at scale.
• Aggregated Hardware Supply: AI Supercloud connects Parasail’s internal GPU fleet with various compute providers, unlocking flexible capacity beyond the availability of a single cloud and quickly bringing new hardware platforms to market with disruptive performance.
• Automatic performance optimization: Parasail replaces manual kernel tuning and performance engineering with an automated system that continuously optimizes inference across the network.

“AI infrastructure is moving beyond single-cloud models,” said Samir Kumar, general partner at Touring Capital. “As inference workloads grow, enterprises need flexibility across hardware, geography, and cost structure. Parasail has built a control layer to enable that. We believe the team combines deep systems expertise with a clear product vision, and is well-positioned to define how modern AI applications are deployed.”

“The main product component of this AI wave is agents, replacing the concept of the world of manually operated applications of the past 30 years,” said Steve Jang, managing partner at Kindred Ventures. “While these agents are directed, they operate autonomously, invoke multiple models at runtime, and require large amounts of tokens. This new world and its developers require powerful customized inference and reinforcement learning capabilities that are flexible, immediate, and reliable. Parasail delivers the first agent-centric inference and training solution that simplifies model and computational complexity in today’s dynamically generative AI market.”

Since its launch in April 2025, Parasail has processed over 500 billion tokens per day, with customers including Elicit, mem0, Gravity, kotoba, and Venice, leading the pack of second-wave inference providers with 30% month-on-month revenue growth.