GPU services provider Neysa and Pipeshift, a startup offering managed AI inference platform for open-source models, have partnered to tap rising demand for AI inference in India as enterprises grapple with increasing AI costs and latency linked to wider adoption.
Karan Kirpalani, chief product officer at Neysa, told ET that India’s inference market requires complex AI ecosystems because of its scale and diversity.
“We are seeing a lot of demand for multi-modal, multilingual AI systems that are specialised and local. The partnership between Pipeshift and Neysa addresses those concerns,” he said.
Kirpalani said there is no definitive estimate of India’s inference market, but industry estimates peg it at $28-30 billion in 2025. The global inference market could be worth about $125 billion in 2025 by conservative estimates.
As AI adoption gathers pace, enterprises are facing rising AI costs and latency issues. Sectors such as healthcare and banking also require sovereign control over data.
The partnership aims to address these concerns by combining Neysa’s GPU architecture with Pipeshift’s inference stack to optimise AI model costs through large-scale deployment of open-source models such as Gemma, Qwen, DeepSeek and Mistral.
The offering is aimed at companies operating at scale. The companies said the technology has already been deployed with AI startups Nurix and Arrowhead AI.
The market, however, is crowded, with cloud providers and startups such as Together AI and Fireworks competing in the space.
Arko Chattopadhyay, cofounder and CEO at Pipeshift, said the company is focusing on reducing real-time latency, as even delays of 200-300 milliseconds are critical for many clients.
The company also plans to expand its infrastructure capabilities to include world and video models.
Another challenge is persuading enterprises to adopt open-source AI models at scale.
“For our part, we need to explain to people that Anthropic, OpenAI and Gemini are not the answer to everything all the time,” Chattopadhyay said.
