Simplismart Raises USD 14 Mn Series B Led by Dallas VC

Simplismart, an AI inference infrastructure company, has raised USD 14 million (INR 124 crore) in a Series B round led by Dallas Venture Capital. NVIDIA, Accel, Shastra VC and Micromax also participated.

The funds will go toward developing its inference platform, expanding cloud-provider and sovereign AI partnerships, and growing in North America and other global markets. They will also support a dynamic GPU scheduling layer that matches AI workloads with available compute across regions and time zones.

Simplismart builds the software layer between AI models and GPU infrastructure. Its modular inference engine lets businesses configure model performance by speed, cost and requirement, while its orchestration platform manages deployment and scaling across cloud and on-premises infrastructure. Its annualised revenue run rate has grown 14x over the past six months. HeyGen, Higgsfield, InVideo, Shiprocket, Apna and Ema run production workloads on the platform.

Customer communications platform Exotel post-trained an open-source model on its own data and moved inference to Simplismart. By fractionalising GPUs, Simplismart cut Exotel’s cost per minute by up to 57% compared with closed-model APIs, and below the cost of self-hosting the same model. Response time fell from 1.1 seconds to under 300 milliseconds.

For cloud providers, Simplismart works with NVIDIA and does not own compute. Its software turns third-party GPU fleets into production AI infrastructure, allowing providers to offer managed AI services under their own brands. Yotta uses the platform for the serverless inference layer of Shakti Studio.

Amritanshu Jain, Co-founder and CEO, Simplismart, said: “With frontier-class open models now readily available, access alone is no longer enough to set a business apart. The advantage comes from how you run them: what each interaction costs, how quickly the application responds, and whether it stays reliable as demand grows.”