Jarvislabs.ai, a company located on the outskirts of Coimbatore, India, is providing affordable AI infrastructure services, specifically GPUs on rent, to budding entrepreneurs and developers. Their mission is to make AI deployment accessible and cost-effective, especially for small companies in cities like Chennai or Bengaluru who may find services from hyperscalers like AWS or Google Cloud Platform to be too expensive. Vishnu Subramanian, the founder and chief of Jarvislabs.ai, stated that their GPU rental service is 60 to 70% cheaper compared to using a hyperscaler.
To keep GPU prices low, Jarvislabs.ai adopts a strategy that involves using alternative GPUs with similar capabilities instead of pursuing expensive options like NVIDIA’s H100s. They currently offer Quadro RTX 5000 and 6000 GPUs, which are as good as older generation A100 GPUs but more affordable. Subramanian explained that the RTX 6000 Ada, for example, provides similar performance to an A100 but at a price of $1.14 compared to $3.
Jarvislabs.ai has managed to obtain GPUs at competitive prices by partnering with a national distributor named RP Tech in India. This partnership allows them to acquire GPUs at a better price compared to the open market, which tends to have higher prices on platforms like Amazon. Despite minimal marketing efforts, the company has gained customers through word of mouth and has attracted notable clients such as Zoho and upGrad.
While Jarvislabs.ai focuses on providing GPU rental services, they aim to be more than just a hardware company. They package their offerings as a comprehensive solution that addresses the challenges of training and deploying AI models. With automation and simplified processes, they make it easier for small companies without dedicated DevOps or AI teams to utilize their services.
As a bootstrapped company, Jarvislabs.ai cannot offer substantial discounts like VC-backed firms. However, they provide additional value to their customers by simplifying the AI deployment process and offering an optimized environment. They are also expanding their offerings by working on a new version of their product that will include support for multiple frameworks beyond PyTorch, TensorFlow, and Fast.ai.
The source of the article is from the blog xn--campiahoy-p6a.es