Technology

Nvidia Backs an Indian Startup That Helps AI Run on Fewer Chips

Martin HollowayPublished 2w ago4 min readBased on 2 sources
Reading level
Nvidia Backs an Indian Startup That Helps AI Run on Fewer Chips
source:simplismart.ai

Nvidia has taken part in a $14 million funding round for Simplismart, a four-year-old startup in Bengaluru, India. The company makes software that helps other businesses run AI models more efficiently, in some cases on fewer GPUs, according to TechCrunch.

GPUs are the specialized chips that do most of the heavy computing for AI.

The round was all-primary. That means the money goes to the company itself, not to existing shareholders who sell their stakes. Dallas Venture Capital led the round. Other investors in the Series B include Accel, Shastra VC and Micromax. The company is valued at about $100 million, including the new money.

Simplismart works in what the industry calls the inference layer. Training is when an AI model learns from data. Inference is when the finished model answers questions from real users. A helpful comparison is a restaurant. Training is writing the recipes, and inference is the busy kitchen filling orders all day. Simplismart's software works in that kitchen.

According to the same report, Simplismart helped AI video-generation platform InVideo cut its GPU serving costs by 56%. Its other customers include HeyGen, Higgsfield and Ema. Yotta, an Indian cloud provider, uses Simplismart's technology to power a serverless AI service inside its Shakti Studio platform. "Serverless" means customers pay for each request and do not manage their own servers.

The company's annualized revenue run rate grew 14-fold over the past six months. Run rate is a way of projecting current revenue over a full year. The figure now sits at an eight-figure sum, which means at least $10 million.

More than half of the company's revenue comes from the United States, even though it has more customers in India.

Nvidia has also been working with Simplismart on technical integrations. These include work around Dynamo, Nvidia's software for serving AI models.

Simplismart's earlier funding was a $7 million Series A, led by Accel. The company announced it in October 2024 on its blog. The new round more than doubles the capital the company has disclosed.

The broader context here is the cost of running AI models, as opposed to building them. A company that builds a product on top of an existing AI model pays an ongoing bill for every answer its users receive, and that bill grows with the user count. A cut like the one reported for InVideo can decide which products are affordable to offer. It can also change how much computing hardware a customer needs to buy or rent.

The revenue mix is worth a closer look. In my view, a company with more customers at home but most of its revenue in the U.S. suggests that businesses willing to pay for this kind of optimization are concentrated in American app developers, at least for now. How that balance shifts will show whether the company's growth depends on one market or on a product that works well across borders.

The Nvidia tie is worth considering on its own terms. Simplismart's work is being built around Dynamo, a platform that many operators already use, which can speed adoption. In my view, the same tie also means the company's plans depend partly on software controlled by a supplier. Customers and investors will likely judge the company by whether its speed gains hold up as the underlying tools change.

What the round makes possible is fairly plain. Cheaper serving lowers the cost of running video, avatar and AI agent tools at large scale. It also gives smaller companies access to computing power that once required a large budget. Whether Simplismart becomes a lasting part of that foundation will depend on sustained performance gains, on customers who stay after their first savings, and on its ability to keep pace with the platforms it relies on.