Nvidia invests $14 million in Indian AI inference startup Simplismart

Optimising around the customer

Simplismart’s software sits between AI models and computing infrastructure. It adjusts deployment and management to balance response times, request volumes and compute use, including on GPUs a customer already has.

The Bengaluru startup offers its platform on its own service or in customers’ cloud and on-premises environments. Its tools also include:

Cloud providers can use the software to build and maintain their own GPU-based AI inference services.

Simplismart says video-generation platform InVideo cut GPU operating costs by 56% using its software. Other customers include video-generation companies HeyGen and Higgsfield, and enterprise AI startup Ema. Indian cloud provider Yotta uses Simplismart technology for a serverless AI inference service on its Shakti Studio platform.

The company says its annualised revenue run rate grew 14-fold over the past six months to an eight-figure amount, co-founder and CEO Amritanshu Jain told TechCrunch. The figure is based on recent customer usage, not signed recurring-payment contracts, and excludes money passed through to infrastructure providers. Jain did not disclose the exact amount. More than half of revenue comes from the US, even though Simplismart has more customers in India.

Nvidia’s two-sided bet

Alongside the investment, Nvidia is working with Simplismart on technical integrations, including ones involving Nvidia’s Dynamo inference software. Simplismart engineers and Nvidia solution architects are involved; Jain said Nvidia is also introducing the startup to potential customers and partners.

That makes the investment more than a bet on a standalone software vendor. Nvidia is backing a company that aims to make inference workloads more efficient, including by helping some customers run them with fewer GPUs. Jain argues that better inference efficiency can still benefit Nvidia by making its hardware more attractive and helping AI applications spread.

Simplismart is competing with inference offerings from Nvidia and major cloud providers, including Amazon Web Services, Google Cloud and Microsoft Azure, as well as startups building tools for faster, cheaper model deployment. Jain says Simplismart’s advantage is for customers with strict service-level agreements and tight control of total cost of ownership: rather than choose between preset options such as fast or cheap inference, they can tune the balance to their needs.

Growth with a caveat

Jain and former Google engineer Devansh Ghatak founded Simplismart in 2022, drawing on their experience building and deploying machine-learning systems. The two met while studying at BITS Pilani. The company now has 58 employees, Jain said.

The new investment will fund US expansion, hiring and further inference-software development. The announcement followed a report by The Economic Times that Simplismart was in talks to raise $40–50 million at a $300–400 million valuation. Jain declined to discuss details, said the company was not actively seeking investment and stated that its Series B round was closed and would not be extended.

I think the most revealing number here is not the 14-fold run-rate growth, but what it leaves out: the figure is usage-based, excludes infrastructure pass-through and has no disclosed dollar amount. That makes it hard to judge how much of the reported momentum belongs to Simplismart’s software economics rather than customer spending on compute. Nvidia’s support may help the startup reach more buyers, but its case will ultimately rest on whether customers can verify the promised balance of performance and cost.