Nvidia Backs Bengaluru Inference Startup Simplismart in $14 Million Series B

Nvidia participated in a $14 million Series B round for Simplismart, a four-year-old Bengaluru startup whose software helps companies run AI models more efficiently, in some cases on fewer GPUs, according to TechCrunch.

The round was all-primary, meaning the new capital goes to the company rather than to existing shareholders. Dallas Venture Capital led it. Other Series B investors include Accel, Shastra VC and Micromax. The company was valued at about $100 million including the new funding.

The product sits in the inference layer, the part of the stack where trained models serve live requests. According to the same report, Simplismart helped AI video-generation platform InVideo cut its GPU serving costs by 56%. Its customers include HeyGen, Higgsfield and Ema. Yotta, the Indian cloud provider, uses Simplismart's technology to power the serverless AI inference service within its Shakti Studio platform.

The company's annualized revenue run rate grew 14-fold over the past six months, to an eight-figure sum. More than half of its revenue comes from the United States, even though it has more customers in India. Nvidia has been working with Simplismart on technical integrations, including around its Dynamo inference software.

Simplismart's earlier $7 million Series A, led by Accel, was announced in October 2024 on the company's blog. That round is the company's previous funding event, and the new round more than doubles its disclosed capital base at the Series B stage.

The broader context here is the economics of serving models rather than training them. For companies building on top of open or licensed models, GPU serving cost is a recurring line item that grows with every user request. A reduction of the size reported for InVideo changes which products can be priced sensibly for consumers, and it changes how much hardware a given customer needs to buy or rent. Software that squeezes more throughput from each accelerator therefore sits directly against the cost curve that application companies face.

The revenue mix deserves a second look. An Indian company with a larger customer count at home but most of its revenue from the U.S. suggests that willingness to pay for inference optimization is concentrated in American application developers, at least for now. How that balance shifts will show whether the company's growth depends on a particular market or on a product that travels across borders without much adaptation.

The Nvidia relationship is also worth weighing on its own terms. The reported technical work around Dynamo, Nvidia's inference-serving software, means Simplismart's stack is being built against a platform that many operators already use. That can accelerate adoption, but it also ties the company's roadmap to a platform owned by a supplier. Investors and customers will likely judge the company on whether its optimizations hold up as the underlying serving frameworks change.

What this round enables is straightforward. Cheaper serving lowers the threshold for running video generation, avatar and agent workloads at scale, and it gives smaller operators access to capacity that previously required large infrastructure budgets. Whether Simplismart becomes a durable part of that layer will depend on the same things that have decided earlier infrastructure companies: sustained performance gains, customer retention once the initial cost savings are absorbed, and the ability to keep pace with the platforms it depends on.