Cloud startup Neysa Networks Pvt. Ltd. today announced that it will raise up to $1.2 billion in financing to build data center infrastructure.

A consortium led by Blackstone will provide up to half the sum in the form of equity funding. According to TechCrunch, the deal is set to buy the investor group a majority stake in Neysa. The Mumbai-based company also plans to raise $600 million in debt.

“This investment positions Neysa to play a meaningful role in advancing AI infrastructure in India and enables businesses and public institutions to deploy AI technologies more effectively as AI adoption accelerates,” said Blackstone Senior Managing Director Ganesh Mani.

Neysa operates a public cloud platform optimized to run artificial intelligence workloads. The platform is reportedly powered by about 2,000 graphics processing units. The most advanced AI accelerator currently offered by Neysa is Nvidia Corp.’s H200, which debuted in 2023 and provides 4 petaflops of performance when processing FP8 data.

The company’s cloud platform also features more general-purpose chips. Customers can rent virtual machines powered by Nvidia’s L40S accelerator, which combines AI-optimized circuits with modules geared towards graphics rendering tasks. Additionally, Neysa offers several instances equipped solely with central processing units.

The company will reportedly use most of its newly raised capital to grow its infrastructure footprint. Neysa plans to increase the number of AI accelerators in its cloud nearly tenfold to 20,000 units. According to Blackstone, that’s equal to about one third number of AI-optimized data center GPUs deployed in India today.

Given that the newest Nvidia graphics card supported by Neysa was released more than two years ago, it’s possible the company’s infrastructure buildout will focus on obtaining more advanced chips. Nvidia debuted its latest GPU, Rubin, early last month. The chip includes 336 billion transistors that can deliver 50 petaflops of performance when processing NVFP4 data.

Neysa may deploy some of the GPUs that it plans to buy in the Hyderabad data center it previewed last April. The facility will cost $1.2 billion, the maximum amount of financing the company expects to access through its newly announced investment. Neysa plans to equip the the data center with up to 25,000 graphics cards.

The company offers its infrastructure services alongside a suite of software tools. Neysa’s platform enables developers to fine-tune, or customize, open-source AI models for specific use cases. A software team can create multiple customized versions of an AI model and use an experiment tracking tool included in the cloud platform to identify the best-performing version.

After the training phase is complete, Neysa helps customers monitor their AI models in production. There’s a built-in observability tool that tracks metrics such as a neural network’s memory usage. Additionally, Neysa provides security controls that make it possible to regulate which user may access a model and how.

The company will reportedly build new workload security, observability and management features using its new funding. It hopes that the engineering investments will help triple its revenue over the next few years. 

Photo: Unsplash