India’s AI future is at an inflection point. With global giants like Reliance, Meta, and Google making bold AI infrastructure bets, the question of sovereignty, compliance, and control has never been more critical. While partnerships democratize access, they also highlight India’s dependence on foreign ecosystems for computing power and models.
In this exclusive conversation, A S Rajgopal, MD & CEO of NxtGen Cloud Technologies, shares how his company is building a sovereign-first AI and cloud backbone for India by balancing GPU dependence, compliance mandates, and the urgent need to transform coders into deep-tech AI specialists.
Reliance has partnered with Meta and Google for AI infrastructure. How sovereign-first clouds will shape India’s AI future, and where does NxtGen’s strategy fit in that landscape?
Reliance’s move with Meta and Google is significant, but it’s a business-oriented alliance rather than a fundamental bet on infrastructure. They are leveraging Meta’s Llama models and Google TPUs to deliver AI services at scale in India. This will democratize access to AI, but does not change the fact that control of core infrastructure and models remains outside India.
Sovereign-first clouds will shape India’s AI future by addressing three non-negotiables:
(1) data residency and compliance with RBI, SEBI, and government norms.
(2) resilience against geopolitical risks in access to computing; and
(3) tailoring to the Indian context and languages without dependency on foreign control.
NxtGen’s strategy fits well here; we are building India’s sovereign AI and cloud backbone. Our approach is OEM-agnostic and GPU-inclusive, running 512× H200 GPUs today on our SAS-F framework, and scaling further with both NVIDIA and AMD. Unlike consumer-facing ventures, we are creating a sovereign platform for enterprises and government where infrastructure, orchestration, and AI agents are under India’s control.
With GPUs becoming the new oil, how do you see India balancing dependence on Nvidia/AMD with building indigenous AI capacity?
GPUs are today’s strategic commodity, the “new oil.” Every major AI initiative, from ChatGPT to DeepSeek, is fueled by GPU access. India cannot escape dependence on NVIDIA and AMD in the short term because these ecosystems are mature and irreplaceable for enterprise-scale AI training today.
However, two things are happening in parallel. First, advancements in CPU technology (from AMD and Intel) are enabling inference for models with up to 10 billion parameters, creating viable alternatives to edge TPUs for many practical use cases. Second, startups like Sambanova and Cerebras are building inference-first hardware that reduces dependence on a single ecosystem.
India’s balancing act will be two-fold: secure long-term GPU supply through partnerships and sovereign capacity (as NxtGen is building), and invest in indigenous innovation in accelerators, compilers, and quantum-safe software layers. Our role is to de-risk India’s AI journey, not by replacing NVIDIA/AMD overnight, but by ensuring India is not hostage to any one vendor or geography.
With your new sovereign, open-agentic AI platform, how do you envision it changing enterprise adoption in India? What’s the core vision behind it, and how will its architecture or capabilities stand out?
Our sovereign, open-agentic AI platform, “M,” is designed around a simple premise: AI should be action-oriented for real-world, sovereign, and open applications.
The vision is to move beyond closed models and chatbots toward agentic workflows that integrate directly into enterprise systems. Imagine an AI that not only drafts a loan approval note but also executes the transaction, updates compliance systems, and informs the customer, all securely, within India’s sovereign boundaries.
Architecturally, M is modular, agentic, and OEM-agnostic. It orchestrates multiple open-source models through our SAS-F framework, optimising for task, cost, and performance. It is deeply multilingual, designed to scale across India’s languages and regulatory regimes. What makes it stand out is the sovereign-first foundation: all data and workflows run on infrastructure within India, on our sovereign cloud, with enterprise-grade SLAs.
For enterprises, this changes adoption in three ways:
Trust – compliance and sovereignty are assured.
Action – AI becomes operational, not just conversational.
Economics – leveraging open-source and sovereign infra brings down cost-per-transaction dramatically.
How do you see upcoming data protection and AI governance laws shaping demand for sovereign cloud and AI platforms in India?
India’s DPDP Act and upcoming AI governance frameworks will be the inflection point. Compliance will no longer be optional; enterprises will be accountable for data residency, consent management, and algorithmic transparency.
Global models and cloud services are not designed with Indian law in mind. Sovereign platforms like ours are built from the ground up to align with these requirements. That means local data residency, quantum-safe encryption pilots for the Government of India, and auditable AI pipelines.
Demand will accelerate in BFSI, government, healthcare, and telecom—the very sectors where compliance and sovereignty are existential, not optional. We see sovereign cloud + AI platforms moving from being a “nice-to-have” narrative to a “license-to-operate” necessity within the next 24 months.
You’re hiring 400 AI engineers. How do you plan to nurture fresh talent into deep-tech specialists when India faces a serious AI skills gap?
India has an abundance of raw talent, but faces a serious AI skills gap in deep-tech engineering. Our strategy to nurture 400 AI engineers focuses on three dimensions:
Partnerships with academia and research – We are already working with IBM Research on Qiskit for quantum, and plan to extend this model to AI engineering programs with leading Indian institutes.
Hands-on problem immersion – Our engineers don’t just learn frameworks; they build real-world use cases, AI agents for IT helpdesk, finance, customer service, manufacturing anomaly detection, and more. By working with live enterprise problems, they mature quickly.
SAS-F framework as a teaching layer – Our patent-pending framework standardises AI solution building. Freshers can learn structured solutioning rather than jumping blindly into model experimentation.
This is how we turn India’s abundant coding pool into deep-tech practitioners. Our commitment is not just to train engineers for NxtGen, but to create an ecosystem of sovereign AI specialists who will power India’s AI future.