Growing alliances between major artificial intelligence firms and global cloud providers are creating structural dependencies that could restrict access to compute and distort pricing for startups, analysts and industry executives said, warning that India must prioritise domestic compute infrastructure to avoid long-term concentration risks.
$165 Billion Lock-in
The concerns come amid a series of deepening partnerships between AI model-makers and hyperscalers. Towards the end of April, Anthropic expanded its strategic ties with both Amazon and Google. Amazon committed a fresh investment of up to $25 billion in the AI firm, while Anthropic agreed to spend more than $100 billion over the next decade on Amazon Web Services (AWS), which will serve as its primary cloud provider and training partner for large-scale AI workloads.
Days later, Anthropic entered a similar arrangement with Google, under which the Alphabet subsidiary plans to invest up to $40 billion in the company and support its growing compute requirements through Google Cloud.
Experts said such arrangements are consolidating control over both foundational models and the infrastructure required to train and deploy them.
“Massive vertical integration and tight partnerships in Big Tech creates monopolistic risks with far reaching implications in general, not just for startups. It gives them undue pricing and bargaining power and allows them to capture even more of the already expensive and supply-constrained hardware right now. This also enables even more tighter proprietary bundling of services,” Kailash Nadh, chief technology officer at Zerodha, told Fe.
Gaurav Parab, principal research analyst at NelsonHall, said the issue was less about outright exclusion and more about long-term dependence. “This isn’t a monopoly in the traditional sense but an ever-increasing dependency on a small number of platforms that control both frontier models as well as compute,” he said, adding that Indian startups were becoming structurally reliant on infrastructure they did not control.
Parasite Risk
Industry executives, however, said startups could reduce exposure through flexible infrastructure strategies. Deepak Dhanak, co-founder and chief operating officer of vibe coding platform Rocket, said firms should avoid dependence on a single cloud provider. “If my infrastructure layer is easily switchable from Azure to Google to AWS, then there is no dependency. In fact, it is good because there is a good amount of competitive pricing pressure that you can exert on them instead,” he said.
Dhanak added that startups building thin wrappers around foundation models faced higher risks. “If I am a parasite and if I have a very shallow wrapper on top of any foundation model, then I’m going to be dependent,” he said. At the same time, he added that such partnerships also simplify access to AI infrastructure and services. “It is a distribution monopoly, not the conventional form of monopolies we find in oil or telecom,” he said.
Most experts said India’s priority should be building domestic compute capacity and strengthening open-source ecosystems rather than attempting to immediately replicate frontier AI models.
“There is no silver bullet solution. We must have our own ecosystem and research capacity,” Nadh said, adding that India should promote open-source ecosystems, interoperability standards and stronger competition regulation.
Parab echoed the view. “Our focus shouldn’t be on immediately trying to replicate frontier models, but on building reliable domestic compute access, supporting open-weight ecosystems, and strengthening middleware and domain-specific AI capabilities,” he said.