It is not easy to build an AI startup in India. While the time is ripe now, getting the funds to build a product for the global market or even a customised solution for the Indian market is extremely difficult. Especially if one is working on the hardware and infrastructure layer, the issue gets maximised. Contrastingly, this is not something that Neysa faced.
Founded by Sharad Sanghi and Anindya Das, Neysa, a Mumbai-based AI acceleration cloud system provider, bagged $20 million in seed funding in April 2024. Cut to October, the team raised Series-A of $30 million, led by NTTVC, Z47, and Nexus Venture Partners.
Investors value the expertise that comes with an experienced founder, and Sanghi exemplified this with his leadership. Neysa is his second venture in the computing space after Netmagic, one of India’s first IT infrastructure startups, which he founded in 1998. NTT later acquired the company, and Sanghi continued working there as the CEO.
Das is currently the chief technology officer, and Karan Kirpalani is the chief product officer. Both, part of the core team members and friends of the founders over the last two decades, discussed with AIM the moat of the company in a world that is filled with data centres, AI infrastructure providers, and Indian AI startups.
With an AI-first philosophy, Kirpalani said that Neya is redefining the landscape of AI-driven solutions with its three flagship offerings, Velocis, Overwatch, and Aegis. However, the most important one right now is Velocis AI.
What’s the offering?
The Velocis AI Cloud is Neysa’s comprehensive platform designed to support the entire AI lifecycle—from training and fine-tuning models to inferencing. “Velocis is built as an end-to-end cloud platform that covers everything from data ingestion to inferencing,” said Kirpalani, highlighting their differentiating factor.
Kirpalani claims that Neysa is the only company in the world that provides an entire AI Platform-as-a-Service (PaaS), which includes data ingestion, database, data engineering, model repositories, MLOps, and leading to inferencing in its upcoming products.
While Google’s Vertex, Microsoft Azure, or AWS Sagemaker are all in the same space, Kirpalani quotes Neysa’s customers as saying that there is a massive pushback for these providers because of their blackbox proprietary platforms and that they are embracing open-source solutions. This allows clients to integrate any tool of their choice on Neysa’s AI PaaS platform.
Velocis embraces an open-source approach, allowing seamless integration with tools like Hugging Face, TensorFlow, and MLFlow. Kirpalani emphasised that the availability of open-source models and the whole pipeline make it stand out from competitors.
On the lines of Databricks or Snowflake, Kirpalani said Neysa’s Velocis, too, provides multiple deployment options, including bare-metal servers, virtual machines, and managed services catering to diverse client needs. These services are similar to Databricks and Snowflake hosted on cloud providers, and finding the moat here for Neysa is not clearly visible.
Sovereign AI is a Moat for Neysa, But so is for Others in India
On the hardware side, Neysa works closely with NVIDIA and provides fractional GPUs for small use cases. This also includes GPUs like L4s, H100s, H200, and several others. “We don’t want to get into the legacy GPUs like A100 that other players are still providing,” Kirpalani said, adding that they don’t want to provide consumer-grade GPUs like RTX and GTX.
When asked about Yotta, one of Neysa’s closest competitors and also a close Indian client of NVIDIA, Kirpalani and Das declined to comment. But Kirpalani added that Indian providers are able to leverage existing infrastructure to provide just bare metal services, while Neysa’s focus is to offer an entire cloud stack that no one is building.
At the same time, Yotta also offers similar services, from training on bare metal hardware to deploying custom models and inference on its Shakti AI Cloud platform.
That said, Kirpalani noted the recent announcement of their Insurance AI Cloud in partnership with Data Science Wizards, offering an end-to-end cloud platform for insurance companies. For one of their clients, Kirpalani said Neysa witnessed an 18% uptake in persistence, which is one of the crucial factors when implementing AI in the insurance sector.
In another example, Kirpalani said that a financial services company that was using Neysa’s offering in their call centre was able to demonstrate a 67% TCO reduction, reaching to INR 4.7 per minute. This was using Meta’s Llama 3.1 on the Velocis AI PaaS platform with NVIDIA H100 GPUs.
Before, the client was using OpenAI on Microsoft Azure, which cost them INR 15 per minute.
Kirpalani also added that the company is in talks with Fractal about building something in the insurance sector for the clients. “We would have to work with Fractal at the sharp end of the stick for their very, very large global clients. The middle layer, which is a little more mass market offering, is our partnership with DSW,” he added.
For the road ahead, Neysa plans to push Aegis, its security platform, which focuses on emerging threats unique to the AI landscape. Aegis provides robust protection measures, ensuring secure AI development and deployment. Neysa plans to expand it into a standalone product by early 2025.
In the future, Neysa wants to launch inference-as-a-service by early 2025, further enhancing its AI lifecycle offerings. “We’re accelerating our timelines to meet market demand,” Kirpalani shared.
The post What’s the Moat of Neysa in the World of Yotta? appeared first on Analytics India Magazine.
