nvidia just backed an indian startup that helps companies run AI with fewer of nvidia’s own GPUs it’s called simplismart, bengaluru-based, $14 million series B with accel. valued at around $100 million their software sits between AI models and the computing infrastructure, and automatically tunes how models are deployed so companies get the right balance of speed, cost, and performance without manually managing it invideo cut their GPU serving costs by 56% using it. revenue grew 14x in six months. more than half comes from the US the obvious question is why nvidia would back software that reduces demand for nvidia chips the founders’ answer: making AI inference cheaper and easier gets more companies to adopt AI in the first place, which ultimately sells more GPUs nvidia is also working with simplismart on technical integrations with its own Dynamo inference software, introducing them to customers and partners so nvidia isn’t just investing, it’s embedding itself into a startup that competes in the same space as nvidia’s own inference products either nvidia genuinely believes efficiency drives more adoption, or they’d rather own the efficiency play than watch someone else take it