Neysa Velocis, Mumbai-based Neysa’s flagship platform, helps scale up AI projects using GPU infrastructure and has a diverse inventory of thousands of GPUs. Indore-based NeevCloud is a recent entrant with plans to deploy 40,000 GPUs, aiming to provide affordable and accessible GPU resources to AI researchers and developers.
Bengaluru-based Jarvis Labs has access to more than 1,000 GPUs through partners and also has its own capacity in the country.
Small companies often offer better local support and quicker response times, factors that are crucial for businesses operating in India, said the people cited earlier.
But setting up and maintaining GPU infrastructure can be expensive, especially for smaller companies. The market is becoming increasingly competitive, with both large and small players vying for market share.
Also Read: Small companies can now chase India’s giant AI dreams
Discover the stories of your interest
However, the growing demand for AI and machine learning in India presents a significant opportunity for small companies to thrive.“Very shortly, we will have a much more recent and a more modern inventory of GPUs within our platform,” said Karan Kirpalani, chief product officer at Neysa, an AI acceleration cloud platform company.A large part of Neysa’s clientele in the private sector is from financial services, insurance, education, media, entertainment, pharma, life sciences and IT/ITeS.
Neysa Velocis offers GPU-based AI services including training, fine-tuning and inferencing, with a focus on cost efficiency and ease of use. It has raised $50 million in funding and collaborates with NTT Data Corporation’s data centres in India. It has onboarded 12 clients in about 10 weeks since the formal launch of its platform in October, with a mix of Indian and global multinationals.
Neysa’s differentiation comes from its intellectual property and cost-conscious design, said Kirpalani, adding that its pricing is competitive, offering return on investment benefits and total cost of ownership reductions of 40-70% compared to hyper scalers. It is also participating in the India AI mission, leveraging partnerships for scale and complexity.
Narendra Sen, founding chief executive of NeevCloud, told ET that the company has built an orchestration engine to provide GPUs on demand. “We can expand in the US and Europe and are continuing to invest in AI data centres in central India, and are going to other states like Chhattisgarh, Raipur, Mumbai and Bengaluru,” he said.
Sen said that these data centres will be capable of holding up to 60,000 GPUs, but the company will ramp up next year with around 2,000 of Nvidia’s H200 and Blackwell GPUs to begin with.
Also Read: Govt eases sustainability push in AI GPU tender after pushback
Jarvis Labs, founded in 2020, initially provided integrated development environments powered by GPUs at affordable prices due to lack of competitors. By 2023-2024, major companies entered the market, prompting Jarvis Labs to explore newer markets like server-less and AI-as-a-Service (AIaaS) models aimed at simplifying the deployment of custom AI models for startups.
By providing application programming interfaces, the company aims to reduce the engineering effort required to build AI pipelines, making it easier for startups to integrate AI solutions.
It had soft launches of these services and partnered with global GPU providers for competitive pricing. The company’s primary market is overseas, and its Indian base is also growing. It aims to cross $1 million in revenue this year, having earned $0.6 million last year with a team of four.
Jarvis Labs’ founding chief executive Vishnu Subramanian told ET, “The company is focusing on server-less and AIaaS models to bridge the gap between application builders and raw GPU providers.”
The company’s GPUs are housed in a tier-3 data centre near Bengaluru and through partners it is able to offer GPUs in Europe, Canada and India.
The company has access to more than 1,000 GPUs through these partnerships, allowing for scalable GPU usage without the need for direct ownership. Vishnu said that the company purchased more than 100 GPUs as a bootstrapped startup. Asked about the customer base, he said the company started with individual users such as researchers and Kagglers.
In recent years companies such as Zoho, NIIT, Upgrad and Thetatech have become customers. Vishnu emphasised the company’s goal of making AI accessible and affordable, despite initially having to build its own infrastructure for want of viable options.
Kirpalani said India’s market requires more hand-holding and packaged solutions for clients to achieve AI objectives. He highlighted the importance of tailored solutions for the domestic market, where specialised local players understand the nuances better. He also said that clients care more about the overall performance and cost of the GPU rather than the specific model.
Also Read: GPU access to power up India’s AI play: experts
