Nvidia invests in Indian startup Simplismart, which helps run AI models with fewer GPUs

The round and its participants

Nvidia has joined the $14 million Series B round of the Indian startup Simplismart, TechCrunch reported. The round was entirely primary: besides Nvidia, it included Accel, Shastra VC and Micromax, with Dallas Venture Capital leading the deal. The transaction valued the four-year-old startup at roughly $100 million including the new funds.

What the product does

Simplismart develops software that sits between AI models and computing infrastructure and adapts their deployment to a client's requirements: response speed, the number of requests processed and compute consumption. The company claims this makes it possible to get more out of an existing GPU fleet and, in some cases, to make do with fewer of them. The startup offers deployment via its own platform or in the client's cloud and on-premises infrastructure, as well as tools for model fine-tuning, autoscaling and monitoring; in addition, cloud providers can build their own AI inference services on top of its software.

Customers and metrics

According to co-founder and CEO Amritanshu Jain, Simplismart's software helped the video generation platform InVideo cut its GPU maintenance costs by 56%. Other named customers include HeyGen, Higgsfield, Ema and the Indian cloud provider Yotta, which uses the technology in the serverless inference service of its Shakti Studio platform. The annual revenue run-rate, Jain said, grew 14-fold over six months to an eight-figure sum; the exact figure is not disclosed, and the metric is based on recent customer usage rather than contracted subscriptions and excludes revenue passed on to infrastructure providers. More than half of the revenue comes from the US, although most customers are in India.

The paradox of the investment and competition

The investment looks paradoxical: the dominant supplier of AI chips is backing software capable of reducing the need for GPUs. Jain counters that more efficient inference ultimately benefits Nvidia, making its hardware more attractive and expanding the use of AI. Beyond the investment, Nvidia cooperates with the startup technically, including integrations with its Dynamo inference software, and helps with introductions to potential customers. The market, however, is competitive: Simplismart competes with Nvidia's own products, the major clouds — AWS, Google Cloud and Microsoft Azure — and a growing number of startups. 'Where we usually win is where people are more tightly constrained by SLAs and TCO,' Jain noted.

Founders and plans

The startup was founded in 2022 by former Oracle engineer Jain and former Google engineer Devansh Ghatak, who met while studying at BITS Pilani; the company has 58 employees. The new funds will go toward expanding its presence in the US, growing the team and developing its inference software.

Contradiction with the report of a larger round

The round announcement followed a report by the Indian newspaper The Economic Times, according to which Simplismart was in talks to raise $40–50 million at a valuation of $300–400 million. Asked about this, Jain declined to comment, saying the startup was not actively fundraising, and added that the Series B round was closed and would not be extended. Thus, the confirmed deal is $14 million at a valuation of about $100 million, while reports of a larger round remain at the level of the newspaper's report without confirmation from the company.