India’s AI firms sidestep Big Tech dependence as 90% adopt multi-cloud strategy: Esya Centre

India’s artificial intelligence ecosystem is showing little sign of the concentration fears troubling global regulators, with companies increasingly diversifying cloud infrastructure, deploying open-source models, and reducing reliance on centralised computing platforms, according to a new report by Esya Centre

The study, based on a survey of 227 firms across the AI value chain, found that nearly 90% of Indian companies either use or plan to adopt hybrid or multi-cloud infrastructure — signalling a shift away from dependence on a single hyperscaler. The report comes amid growing international scrutiny over the dominance of large cloud providers and foundation model companies in the AI economy.

Indian firms, however, appear to be taking a more distributed approach. Nearly 62% of respondents said they combine services from smaller cloud vendors with in-house data centres, while more than 80% reported operating without exclusivity clauses in their cloud contracts.

The findings also suggest intensifying competition among infrastructure providers. Almost 99% of surveyed firms said compute costs have declined, with companies citing vendor flexibility and cost optimisation as the primary reasons for adopting multi-cloud strategies.

Performance overtakes pricing in cloud decisions

Contrary to assumptions that AI firms are driven largely by cost arbitrage, the report found performance and security have emerged as the biggest determinants in cloud-provider selection. Nearly three-fourths of respondents rated these factors as “extremely important”, ahead of pricing considerations.

Scalability and service breadth were also ranked highly, underscoring how enterprises are prioritising reliability and deployment flexibility as AI workloads expand.

The report argues that these trends point to a competitive and evolving infrastructure market rather than one dominated by entrenched incumbents.

Open-source AI gains momentum

India’s AI ecosystem is also increasingly leaning on open-source technologies. Around 83% of surveyed organisations said they use open-source foundation models, while 67% rely on proprietary models and 63% deploy custom-trained systems.

Companies are also adopting a mix-and-match approach to model deployment. Three in five firms said they use multiple models within the same category, while 81% combine both small and large language models depending on application requirements.

At the application layer, the report noted that India’s generative AI startup base expanded 3.6 times between the first half of 2023 and the first half of 2024.

Skills shortage remains the biggest bottleneck

Despite rapid growth, firms identified poor-quality datasets, limited access to government data, copyright ambiguity, and uncertainty around the Digital Personal Data Protection Act (DPDPA), 2023, as major operational hurdles.

“The survey data paints a picture of an AI ecosystem that is far more dynamic and competitive than many assume,” said Meghna Bal, Director at Esya Centre.

The report added that the most significant barrier to scaling AI adoption in India is not market concentration or capital access, but the shortage of skilled AI talent.