Nvidia backs Indian startup helping companies run AI models with fewer GPUs

Companies can reduce their AI inference costs by deploying AI models more efficiently.

Nvidia has invested in Simplismart, an Indian startup focused on optimizing AI model deployment to reduce GPU usage. The $14 million Series B funding round, which valued Simplismart at approximately $100 million, saw participation from Accel, Shastra VC, and Micromax, with Dallas Venture Capital leading. Simplismart's software aims to tailor AI model performance to specific business needs by optimizing factors like response speed, request handling, and computing power consumption. This efficiency allows companies to achieve desired AI outcomes with potentially fewer GPUs, addressing the challenge of high computing costs in the AI boom. The startup claims to have helped clients like InVideo reduce GPU serving costs by 56%. Simplismart's annualized revenue run rate has grown significantly in the last six months, with over half of its revenue generated from the U.S. market. Nvidia's involvement includes technical integrations and introductions to potential customers and partners.

The investment in Simplismart, despite Nvidia's primary role as a GPU supplier, signals a strategic shift towards software-driven efficiency in the AI hardware market. Nvidia's backing suggests a recognition that optimizing inference performance and total cost of ownership, rather than simply selling more hardware, is becoming crucial for broader AI adoption. This could lead to increased competition in the AI software optimization space, potentially benefiting end-users through more cost-effective AI solutions.

Simplismart's success in enabling companies like InVideo to slash GPU costs by over half highlights a growing demand for sophisticated inference management tools. As AI models become more complex and compute-intensive, the ability to fine-tune performance and cost will be a key differentiator. Businesses looking to scale their AI initiatives will likely seek solutions that maximize the utility of their existing hardware investments, making software platforms that bridge the gap between models and infrastructure increasingly valuable.