Neysa
Verified CompanyThe home for your AI — build, scale and govern AI in production
The man who built India's largest data centre company did it again, for AI. Founded in 2023, Neysa now runs 20,000+ GPUs across Mumbai and Hyderabad — close to 30% of India's high-end AI compute — backed by a Blackstone-led raise of up to $1.2 billion.
About Neysa
Sharad Sanghi has already built one piece of Indian internet infrastructure. Netmagic, the company he founded in 1998, became India's largest data centre business before NTT acquired it. In 2023 he started again, with Anindya Das, on the assumption that the next scarce resource would not be racks and bandwidth but GPUs. Neysa is what they built: an AI-focused cloud — a 'neocloud', in the industry's own word — headquartered in Mumbai. The premise is narrow and deliberate. It does not try to be a general-purpose cloud. It provides the compute, the platform to build and operate on, and the engineering team that keeps AI workloads running in production, which is usually where AI projects actually fail. The product suite is called Neysa Velocis. It runs across three layers: GPU-as-a-Service for raw compute, AI Platform-as-a-Service for building and orchestrating models, and Inference-as-a-Service for running them in production. Around that sit orchestration and MLOps tooling, real-time GPU monitoring, AI infrastructure security including a product called LLM Shield, an AI catalog and marketplace, and a startup programme. The scale arrived quickly. Neysa is deploying more than 20,000 GPUs across hubs in Mumbai and Hyderabad — by reported estimates close to 30% of all high-end AI compute capacity in India. That number is the reason the company matters beyond its own customers: a country that wants to train and serve its own models needs somewhere domestic to do it. The funding tracks that ambition. A $20 million seed round in April 2024 was led by Matrix Partners India, now Z47, with Nexus Venture Partners and NTTVC — Sanghi's former acquirer backing him again. A $30 million Series A followed in October 2024, co-led by the same investors. Then, in February 2026, private equity funds affiliated with Blackstone and a group of co-investors signed definitive agreements enabling a capital raise of up to $1.2 billion: up to $600 million of equity, against which Neysa intends to raise a further $600 million in debt. Most of it goes into GPU clusters — compute, networking and storage — with a smaller share into R&D and the orchestration, observability and security software. The customer work is more grounded than the headline numbers suggest. The Indian Institute of Science has fine-tuned an open-weight vision model on Neysa. ITQ has reported cutting its AI costs by 40% by moving onto the platform. Innoviti is a published case study. The company has partnered with Pipeshift on real-time inference, and works across technical education and research, AI-native startups, banking and financial services, insurance, manufacturing, and e-commerce and retail. For a company founded in 2023, the trajectory is unusual — but so is the founder's second-time-around advantage. Sanghi spent two decades learning what it costs to put physical infrastructure in Indian cities and keep it running. Neysa is that lesson applied to a different kind of machine. Sources: neysa.ai (product suite, solution verticals, customer case studies and partnerships, accessed August 2026); Blackstone press release and TechCrunch reporting on the February 2026 financing; seed and Series A details as reported by the Indian startup press. Images: Neysa, reproduced from the company's own website for identification.
Our Services
GPU-as-a-Service
Raw accelerated compute, provisioned on demand from Neysa's own clusters in Mumbai and Hyderabad rather than rented from a hyperscaler abroad — which matters for latency, cost and, increasingly, for where an organisation is allowed to keep its data.
AI Platform-as-a-Service
The layer above the metal: environments for building, fine-tuning and orchestrating models, so teams are not assembling their own toolchain from scratch before they can train anything.
Inference-as-a-Service
Running trained models in production at predictable cost — the part of the AI lifecycle that never ends and where most of the spend eventually sits. Neysa has partnered with Pipeshift to offer real-time inference.
Orchestration, MLOps and Monitoring
Workload orchestration, MLOps pipelines and real-time GPU monitoring, so teams can see utilisation and cost as they happen rather than discovering both at the end of the month.
AI Infrastructure Security and LLM Shield
Security built for AI workloads specifically, including LLM Shield, aimed at the failure modes that conventional infrastructure security was never designed to catch.
AI Catalog, Marketplace and Startup Programme
A catalogue of ready models and a marketplace for AI services, plus a dedicated programme for startups — the on-ramp for teams that need compute before they have revenue.
Key Highlights
Our Portfolio
Neysa Velocis
Orchestration & MLOps
LLM Shield
Enterprise Case Studies
Our Clients
Certifications & Partnerships
Recent News & Announcements
Blackstone-affiliated funds and co-investors sign definitive agreements enabling a capital raise of up to $1.2 billion — up to $600 million equity, with $600 million of debt intended alongside
Neysa and Pipeshift launch real-time inference on the Velocis platform
IISc fine-tunes an open-weight vision model on Neysa; ITQ reports cutting AI costs by 40% after moving to the platform
Latest Articles
Alana Evans Net Worth in 2026 – Income Sources, Career & Earnings
Amy Lee Net Worth 2026: Income Sources, Evanescence Career & Personal Life
Abz Love Net Worth 2026: Five Star’s Career, TV Comeback & Income Breakdown
