India's Most Celebrated AI Startup Hits a Strategic Crossroads
Krutrim, the Bengaluru-based artificial intelligence company founded by Ola's Bhavish Aggarwal, achieved unicorn status in early 2024 in record time — becoming India's first generative AI unicorn. Eighteen months later, the company is quietly rewriting its strategy in ways that raise difficult questions about the economics of building large-scale AI models in India. According to reporting by TechCrunch, Krutrim is shifting its primary focus from AI model development to cloud services, following a business overhaul in late 2025 that included pausing chip design efforts and stepping back from competing with the largest global foundation models.
The Brutal Economics of Foundation Model Building
Krutrim's pivot reflects a reckoning unfolding across the global AI landscape. Building and maintaining competitive large language models requires compute resources costing hundreds of crore rupees per training run — and the competitive bar keeps rising. OpenAI, Google, Meta, and Anthropic are collectively spending hundreds of billions of dollars on AI infrastructure in 2026, with each new model generation requiring exponentially more compute than its predecessor.
For an Indian startup, matching this compute intensity is structurally difficult. India's domestic GPU supply chain remains constrained, cloud compute costs in India are higher than in the US on a per-GPU-hour basis, and the talent pool for frontier ML research remains concentrated in a small number of institutions.
What the Cloud Pivot Actually Means
Moving toward cloud services does not mean Krutrim is abandoning AI entirely. The company appears to be repositioning as an Indian cloud infrastructure and AI services provider — helping enterprises access and deploy AI capabilities without building models from scratch. India's enterprise cloud market is growing at approximately 30% annually, and demand for AI-integrated cloud services from Indian businesses is accelerating rapidly. This also aligns Krutrim more directly with the government's IndiaAI Mission, which has allocated Rs 10,372 crore to build domestic AI infrastructure and compute capacity.
What This Means for India's AI Ecosystem
India's startup funding ecosystem has seen a 93% drop in AI-specific investment compared to 2025, even as overall startup funding of $7.36 billion across 743 equity rounds shows the broader ecosystem remains active. The companies gaining traction in India are those applying AI to domain-specific problems in healthcare, agri-tech, financial services, and edtech — rather than attempting to build general-purpose AI models competing on global benchmarks.
Bhavish Aggarwal's Bigger Picture
Aggarwal remains one of India's most ambitious and polarising founders. His Ola Electric venture, which went public in 2024, has faced scrutiny over quality and service network issues but remains the dominant player in India's electric two-wheeler market. Krutrim's pivot may ultimately prove to be the right call — a recognition that building world-class cloud infrastructure for the Indian enterprise market is a clearer path to a large, sustainable business than competing in the global AI model arms race with a fraction of the resources available to American and Chinese competitors. India's AI opportunity in 2026 is vast, but it runs through application and deployment, not model training.
