Nvidia backs Bengaluru AI startup Simplismart in $14 million round

Simplismart, a Bengaluru startup that makes software for running artificial intelligence models, has raised $14 million in a Series B round with Nvidia among its investors, the company said on Tuesday.

Dallas Venture Capital led the round, and Nvidia, Accel, Shastra VC and Micromax also took part, according to the company’s announcement. Dallas Venture Capital invested close to $5 million, co-founder and chief executive Amritanshu Jain told Inc42. The deal values the company at about $100 million including the new money, TechCrunch and Entrackr reported. Simplismart raised $7 million in a Series A led by Accel in October 2024.

The company works on inference, the stage at which a trained model answers requests. Its software sits between AI models and the graphics processors they run on, and sets how each workload is deployed to meet a customer’s targets for speed, cost and capacity. Simplismart does not own computing hardware. Businesses can use its platform, or run it on their own cloud or on-premises servers, and cloud providers can use it to sell managed AI services under their own brands. In India, it powers the serverless inference layer of Yotta’s Shakti Studio platform.

Simplismart said its annualised revenue run rate had grown 14 times over the past six months. Jain told TechCrunch the run rate was now an eight-figure sum, based on recent customer usage rather than contracted revenue, but did not give the exact figure. Customers include HeyGen, Higgsfield, InVideo, Shiprocket, Apna and Ema.

The company gave one example of the savings it promises. Customer communications firm Exotel trained an open-source model on its own data for its voice agents and moved inference to Simplismart. By giving workloads a fraction of a GPU rather than a whole one, Simplismart said, it brought Exotel’s cost to ₹0.90–1.30 a minute, against ₹2.10 for a closed-model API, a reduction of up to 57 percent, and cut response times to under 300 milliseconds from 1.1 seconds, as Business Standard also reported. The figures are the company’s own.

Nvidia’s engineers have worked with Simplismart on technical integrations, including around its Dynamo inference software, and the chipmaker has introduced the startup to potential customers, Jain told TechCrunch. Asked why the leading supplier of AI chips would back software that can let some customers use fewer of them, Jain said more efficient inference could encourage wider use of AI and make Nvidia’s hardware more attractive. “Nobody actually needs the fastest or the cheapest inference,” he said, adding that companies need speeds and costs that suit their applications.

The market is crowded. Simplismart competes with inference offerings from Nvidia itself, from Amazon Web Services, Google Cloud and Microsoft Azure, and from a growing number of startups.

About 55 percent of Simplismart’s revenue comes from the United States and the rest from India, Jain told Inc42. The company plans to use the new money to expand in North America, deepen partnerships with cloud providers and sovereign AI programmes, and build a scheduling layer that matches AI workloads with spare GPU capacity across regions. Its co-founder and chief technology officer, Devansh Ghatak, said the layer would help cloud providers make better use of their hardware.

Jain, a former Oracle engineer, and Ghatak, a former Google engineer, founded Simplismart in 2022 after meeting at BITS Pilani. The company has 58 employees. Jain told TechCrunch the Series B was closed and would not be extended, and declined to comment on an earlier Economic Times report that the company was in talks to raise $40 million to $50 million.

This report is not investment advice.