Sharad Sanghi, Founder and CEO of Neysa and the entrepreneur who earlier built Netmagic into India's largest data center company, sits down with Aakrit Vaish and Pratyush Choudhury at Mumbai Tech Week to break down the real state of AI compute in India. Neysa runs an AI cloud acceleration platform offering bare metal clusters, platform-as-a-service, and model-as-a-service, and Sharad makes a sharp case that India's compute challenge isn't supply but cost and capacity. He walks through how Neysa differentiates from hyperscalers like AWS and Google, where the GPU numbers actually stand, and what founders can realistically get today.
Timestamps:
0:00 Intro: Mumbai Tech Week and what Neysa does
2:06 The three services: bare metal, platform, and model-as-a-service
2:48 Bare metal vs platform vs model, explained
3:58 Inference-as-a-service and the workspace offerings
5:29 How Neysa differentiates from AWS and Google
6:15 Explaining the difference to a customer: the private bank example
8:00 Who's using Neysa: enterprise, startups, government, research
10:24 Why global companies want to come to India
11:19 India's real compute problem: cost, not supply
13:15 The GPU math: 60,000 today, 2 million in five years
14:28 How the Neysa team is structured across cities
16:15 Where a startup should start, and the startup program
17:32 Max capacity: from a few hundred to 1,000 GPUs
20:20 Why large-capacity buyers should commit long-term
If you're a founder, builder or operator thinking about AI infrastructure, GPUs, or building at scale in India, this one's for you.
@sharad_sanghi @aakrit @177pc