India's First GenAI Unicorn Just Changed Its Business Model
When Krutrim became India's first GenAI unicorn in early 2024, it carried enormous symbolic weight — an Indian company building indigenous AI models in India, for India. The ambition was bold. The subsequent reality has been instructive.
Krutrim has pivoted. After months of relative quiet on the product development front, the company announced a strategic shift away from developing proprietary large language models and toward cloud infrastructure and AI-as-a-service offerings. In its FY26 results, Krutrim reported revenues of approximately Rs 3 billion — a threefold increase from the previous year — along with its first annual net profit and operating margins exceeding 10%.
Why Building Foundation Models Is So Hard
Krutrim's pivot reflects a reality increasingly clear across the global AI landscape: building frontier AI models requires compute, data, and talent at a scale accessible only to the wealthiest technology companies. OpenAI, Google DeepMind, Anthropic, and Meta spend tens of billions of dollars annually on model development. The gap between what these companies can deploy and what even well-funded startups can match has widened, not narrowed, over the past two years.
For Indian AI startups, building at the frontier is prohibitively expensive. Building on top of US-based models is commercially viable but creates dependency and data sovereignty concerns. Building specialized models in Indian languages, for Indian enterprise workflows, for Indian regulatory environments, is a middle path that several companies are exploring with genuine traction.
Neysa: A Different Kind of AI Unicorn
Neysa, an AI acceleration cloud platform, joined the unicorn club in February 2026 after raising $600 million in a Series B round led by Blackstone, valuing it at $1.4 billion. Neysa's model is cloud infrastructure for AI workloads — providing the GPU compute, networking, and software stack that AI companies need to train and deploy models at scale. For enterprises concerned about data sovereignty, a domestic Indian AI infrastructure provider addresses both a commercial gap and a strategic need.
The Broader Ecosystem: $7.62 Billion Raised and Counting
Through May 2026, Indian startups have raised $7.62 billion across 759 equity funding rounds. Skyroot Aerospace entered the unicorn club in May 2026 after raising $60 million in a Series C round, becoming India's first launch vehicle company to reach unicorn status. The AI startup segment specifically has 482 funded companies that have collectively raised $3.4 billion in venture and private equity. Strong clusters exist in AI for healthcare diagnostics, AI for agriculture, AI-powered legal tech, and AI-enabled financial services.
Anthropic Bets on India
Anthropic appointed Sangeeta Bavi as its lead for startups in India, signaling that the company sees India not just as a consumer market for its API, but as a source of the next generation of AI-native companies. For Indian founders, the increasing presence of global AI infrastructure providers building developer ecosystems in India is both validation and an acceleration of available resources.
The Road Ahead
India's AI startup ecosystem in 2026 is something potentially more durable than the frontier AI story it once appeared to promise: a pragmatic, commercially grounded ecosystem finding genuine product-market fit in AI applications, infrastructure, and services. Krutrim's pivot is not a failure; it is a recalibration toward sustainable value creation. That is how lasting technology ecosystems are built.
