Nvidia Backs Indian AI Efficiency Startup Simplismart
Original: Nvidia backs Indian startup helping companies run AI models with fewer GPUs
Why This Matters
GPU efficiency software is emerging as a key cost lever as AI inference spending scales.
Nvidia joined a $14M Series B in Bengaluru's Simplismart, valuing it at ~$100M. The startup's software optimizes AI model deployment to reduce GPU usage, with revenue growing 14x in six months.
Simplismart, a four-year-old Bengaluru startup, has closed a $14 million Series B led by Dallas Venture Capital, with Nvidia, Accel, Shastra VC, and Micromax participating. The round values the company at roughly $100 million. Simplismart's software sits between AI models and underlying compute infrastructure, automatically tuning deployment configurations—response latency, throughput, and resource consumption—so companies can squeeze more from existing GPU stacks rather than buying more hardware. CEO Amritanshu Jain put it plainly: "Nobody actually needs the fastest or the cheapest inference." The pitch is that workloads need inference tuned to their specific cost-and-speed trade-offs. Clients include AI video platforms InVideo, HeyGen, and Higgsfield. InVideo reportedly cut GPU serving costs by 56%. Indian cloud provider Yotta powers the serverless inference layer of its Shakti Studio platform using Simplismart's technology. The startup operates both as a managed platform and as embedded software for cloud providers building their own AI inference services. Annualized revenue run rate has grown 14-fold over the past six months to an undisclosed eight-figure number, based on recent usage rather than contracted recurring revenue, and excluding infrastructure pass-through costs.