Google launches Market Access Program to scale Indian AI startups globally with new AI models

Google Launches Market Access Programme to Scale Indian AI startups | AIM

Google also extended the MedGemma 1.5 and FunctionGemma open-source models for Indian AI health startups. Google has announced a new market access initiative and two open-source AI models to support Indian startups from early-stage development to global scale at the Google AI Startups Conclave held in the national capital. Google Market Access Program aims to help Indian AI startups transition from pilot projects to long-term enterprise contracts. "Indian startups are building serious deep technology and solving population-scale problems with AI," said Preeti Lobana, VP and country manager for India at Google, in a statement. She said that although the journey from labs to prototypes has improved over the past few years, many startups continue to struggle with scaling, a gap the Google Market Access Program seeks to address. The programme targets AI-first startups that have moved beyond the prototype stage and are preparing to scale. It will focus on enterprise readiness through structured training on global enterprise sales, pricing, and buyer behaviour, alongside facilitated introductions to Google's global network of CIOs and CXOs. The initiative also includes international immersion programmes in partnership with ecosystem organisations, including TiE Silicon Valley and Alteus. Applications for the programme are now open. Google also recently announced new additions to its Gemma open model family to support healthcare and on-device AI development. One of the releases, MedGemma 1.5, is a 4-billion-parameter open-source model for medical AI applications. It supports high-dimensional medical imaging workflows, including CT and MRI scans, whole-slide histopathology, longitudinal chest X-ray analysis, anatomical localisation, and extraction of information from medical lab reports. The model builds on Google's Health AI Developer Foundations programme and follows its collaboration with All India Institute of Medical Sciences, which is using MedGemma to develop health foundation models as part of the country's Digital Public Infrastructure. Google also introduced FunctionGemma, a lightweight variant of the Gemma 3 270M model optimised for function calling and on-device AI agents. The model enables applications to convert natural language commands into executable actions locally, allowing AI systems to operate with low latency, limited connectivity, and enhanced user privacy. FunctionGemma can be fine-tuned using tools such as Hugging Face Transformers, Keras, and NVIDIA NeMo, and deployed across environments including LiteRT-LM, vLLM, Llama.cpp, and Vertex AI. Google said these efforts complement its ongoing investments in India's AI infrastructure, including the Global AI Hub in Visakhapatnam, which provides a one-gigawatt compute foundation powered by green energy and Google's AI chips. The company also reiterated its focus on data availability, citing progress on Project Vaani in collaboration with the Indian Institute of Science. The initiative has released more than 27,000 hours of speech data across over 100 Indic languages through the government's Bhashini platform.

Google launches Market Access Program to help Indian AI startups scale globally

Google is boosting India's AI startups. A new programme helps them reach global markets. Indian AI companies are now building real products for healthcare, agriculture, and education. Compute costs are falling, allowing more focus on innovation. India's AI market is projected to reach $126 billion by 2030. India's AI startup ecosystem is moving decisively from pilots to production, with companies increasingly building real-world, revenue-ready products across healthcare, agriculture and education, as Google on Thursday announced new initiatives to help Indian AI-first startups scale globally. At the Google AI Startups Conclave, the company unveiled the Google Market Access Program to help startups bridge the gap between successful pilots and repeatable enterprise adoption. ET Budget Survey: Tell us your wishlist "If you solve for India, you build for the world," Google said, signalling a sharper focus on helping Indian startups reach global markets faster. Also Read: Google and Apple enter into multi-year AI deal for Gemini modelsThe newly launched Market Access Program is aimed at AI startups that have working products but struggle with enterprise readiness, trust and go-to-market execution. The programme will offer global selling expertise, access to Google's enterprise network and immersion in key international markets. Applications are now open. Google said the initiative is designed to support startups at the critical scale-up stage, where access and credibility often determine long-term success. Google also reiterated its full-stack approach to startup support, spanning infrastructure, models, safety, skilling and capital. On infrastructure, the company said Indian startups can build on the same cloud and AI stack that powers Google globally, backed by clean energy and large-scale data centre capacity. It flagged its upcoming Global AI Hub in Visakhapatnam, a 1-gigawatt facility positioned as a long-term anchor for India's AI economy. Startups will also get access to Google's most advanced AI models, including Gemini for complex reasoning and Gemma for open research, alongside Private AI Compute, which combines cloud-based AI with on-device security to keep sensitive data user-controlled. To support deep-tech innovation, Google announced new additions to its open Gemma model family, focused on healthcare and agent-based systems -- two areas seeing rapid adoption in India. MedGemma 1.5, an open 4-billion-parameter model, is designed for population-scale healthcare AI and supports complex medical imaging workflows, including CT and MRI scans, histopathology and lab report analysis. The launch builds on Google's collaboration with AIIMS, which is using MedGemma to develop India's Health Foundation Models as part of the country's digital public infrastructure. Google also introduced FunctionGemma, a lightweight model tuned for function calling, enabling startups to build low-latency, on-device AI agents that can work even on low-end devices and without continuous internet access. Also Read: Google to develop, manufacture smartphones in Vietnam, Nikkei Asia says Alongside the conclave, Inc42 released the 'Bharat AI Startups Report 2026', supported by Google, projecting India's AI market to reach $126 billion by 2030. Nearly 47% of enterprises are already moving AI use cases from pilot to production, the report said. Falling compute costs and public digital infrastructure are allowing founders to shift spending from infrastructure to product innovation. India's complexity -- from linguistic diversity to uneven connectivity -- is increasingly being seen as a strength, with "Bharat-tested" AI emerging as a global benchmark for resilience. Trust is also becoming a key differentiator, with startups that embed safety, privacy and security by design more likely to win long-term enterprise contracts. Google pointed to startups such as Cloudphysician, which has reduced ICU mortality rates by 40%, and Rocket Learning, which is personalising education at scale, as examples of outcome-led AI adoption. (You can now subscribe to our Economic Times WhatsApp channel)