Krutrim, the Bengaluru-based startup recognized as India’s first generative AI unicorn, is undergoing a significant strategic shift. After months of reduced public visibility and a pause in its ambitious chip development efforts, the company is pivoting its primary focus toward providing AI cloud services. This transition follows a major business restructuring in late 2025, which involved reallocating internal capital and personnel to better align with the current economic realities of the high-cost AI sector.

The shift comes as Krutrim faces mounting pressure from domestic competitors like Sarvam, which has recently maintained a high profile through active participation in major industry events and the release of new open-source models. Krutrim’s move away from its initial goal of building large-scale base models follows a turbulent period that included multiple rounds of layoffs affecting over 200 employees and the removal of its Kruti AI assistant app from digital marketplaces earlier this year.

Despite these challenges, the company reports a positive financial trajectory, claiming approximately ₹3 billion in revenue for the 2026 fiscal year and achieving its first annual net profit. However, analysts remain cautious regarding these figures, noting that a significant portion of the startup’s revenue historically stems from its parent organization, Ola, rather than external enterprise clients. Currently, Krutrim asserts that it is seeing increased demand for its cloud infrastructure, with its existing GPU compute capacity largely committed to clients across the healthcare, telecommunications, and financial services sectors.

Industry experts suggest that while prioritizing cloud infrastructure is a commercially pragmatic decision, the company still faces the challenge of proving the long-term sustainability of its business model. As the broader AI landscape in India continues to evolve, Krutrim’s ability to compete with global tech giants and local rivals will likely depend on its capacity to provide scalable, reliable infrastructure rather than relying on the high-stakes development of standalone AI models.