Krutrim abandons AI model ambitions, pivots to cloud services after leadership exodus

India's First GenAI Unicorn Faces Strategic Pivot

Krutrim, founded by Bhavish Aggarwal and celebrated as India's first GenAI unicorn, is making a dramatic strategic pivot from AI model development to AI cloud services after months of operational challenges and restructuring1

1

.

The Bengaluru-based startup announced Tuesday that it is moving toward cloud services following a major restructuring in late 2025 that involved reallocating capital and talent, and pausing chip design efforts. This shift marks a significant retreat from Krutrim's original ambition to build domestic alternatives to models from companies like OpenAI, Google, and Anthropic.

Leadership Exits and Workforce Reductions Signal Deeper Troubles

The company has experienced severe leadership exits and workforce reductions over the past year. Three key AI leaders—Chandra Khatri, the founding head of AI based in Silicon Valley, Sunit S, who headed design, and Raj Kiran, who led the AI research group—all departed in the past quarter2

2

. Additionally, close to 10 executives from Krutrim's semiconductor arm, including Sanjeeb Ghosh, Vyasa Maharshi Grandhi, and Vishnu KGJ, have left in the last six months2

2

. More than 200 roles were cut across multiple rounds, with over 60 linguistics staff laid off in early January, eliminating the last of the company's data annotation team for its AI models2

2

.

Full-Stack AI Ambitions Collapse Amid Funding Constraints

Krutrim raised $50 million at a $1 billion valuation in January 2024, reflecting early investor enthusiasm for India's homegrown AI ambitions1

1

. Aggarwal, who also leads ride-hailing firm Ola and EV maker Ola Electric, made a ₹10,000 crore commitment to develop Krutrim in February 2025, with plans for a full-stack AI platform including large language models for language, speech and vision, semiconductors, and cloud infrastructure2

2

. However, funding dried up, forcing the company to pause work on AI models and semiconductors2

2

. The Bodhi 1 chip, expected for release in 2026, has been scrapped, with chips procured for training models now reportedly being used to run simulations for Ola Electric scooters2

2

. The company also discontinued its AI agentic platform Kruti, pulling the AI assistant app from app stores in April1

1

.

Revenue Claims and Cloud Service Focus Raise Questions

Krutrim claims it generated about ₹3 billion (around $31.52 million) in revenue in financial year 2026, a threefold increase from a year earlier, along with its first annual net profit and margins exceeding 10%1

1

. However, the startup did not disclose how much came from external customers versus parent company Ola's ecosystem, with earlier reports indicating about 90% of FY25 revenue came from group companies1

1

. Sanchit Vir Gogia, chief analyst at Greyhound Research, said the move toward cloud was commercially sensible but cautioned that "the standard of proof must rise with the claim"1

1

. The company now says it has more than 25 enterprise customers across telecom, financial services, and healthcare, with most of its GPU compute capacity already committed to external workloads1

1

.

Competitive Landscape Shifts as Rivals Advance

The move follows a period of limited public activity from Krutrim, with its last post on X dating back to December1

1

. The startup did not appear in any sessions at India's AI Impact Summit in New Delhi, where global players such as Anthropic, Google, and OpenAI participated1

1

. In contrast, rival Sarvam participated in multiple sessions at the six-day AI event, showcasing new open-source models, hardware developments, and commercial partnerships, including a recent one with space-tech firm Pixxel to develop an AI-driven orbital data center1

1

. As Gogia notes, cloud infrastructure may be the more viable near-term play in India's AI market, even as the longer-term ambition of building competitive models persists1

1

. The shift reflects broader challenges facing AI startups attempting to compete with well-funded global players while navigating the expensive economics of building large-scale AI systems.