Simplismart raises $14 Mn to sell the software layer between AI models and GPUs

Simplismart, a Bengaluru-based AI infrastructure startup, has raised $14 Mn in a Series B round led by Dallas Venture Capital, with participation from Nvidia, Accel, Shastra VC, Micromax and angel investors. Dallas Venture Capital put in close to $5 Mn, according to Inc42 and the investor's own announcement.

The company was founded in 2022 by Amritanshu Jain and Devansh Ghatak. It builds what it calls the software layer between AI models and GPU infrastructure: an inference engine that lets a business tune how a model runs for speed, cost and performance, and an orchestration platform that handles deployment and scaling across cloud and on-premises machines.

It sells three ways. Shared endpoints are billed on tokens used. Dedicated endpoints rent GPU capacity bundled with Simplismart's optimisation software. A bring-your-own-compute option runs on the customer's own hardware, which Inc42 reports is aimed at banks, insurers and other regulated businesses that need to keep data in house. Cloud providers can also use the software to offer managed AI services under their own brands; Inc42 reports Yotta uses it for the serverless inference layer of its Shakti Studio platform.

Jain told Inc42 that about 55% of revenue comes from the US and the rest from India, and the company says its annualised revenue run rate grew 14X in six months. Named customers include HeyGen, InVideo, Shiprocket, Apna and Ema. TechCrunch reported that the round values the company at about $100 Mn including the new money, and Entrackr gave the same post-money figure.

The new capital goes into the inference engine, a scheduling layer that matches workloads to spare GPU capacity across regions, cloud partnerships and expansion in North America. Simplismart previously raised a $7 Mn Series A led by Accel in October 2024.