Built on AMD Instinct MI455X GPUs and co-developed with TCS’s HyperVault subsidiary, Helios arrives as India’s AI infrastructure market heats up and as enterprises begin looking seriously for options beyond a single chip architecture.
AMD and Tata Consultancy Services have announced a significant expansion of their strategic collaboration, bringing AMD’s Helios rack-scale AI platform to India through TCS’s data centre subsidiary HyperVault AI Data Center Limited. The partnership will support a 200 MW AI infrastructure deployment designed to serve enterprises, hyperscalers, and AI companies operating in India, including those building under the country’s sovereign AI initiatives.
The timing is deliberate. India’s AI infrastructure market has seen a rush of multi-billion dollar commitments in recent weeks, most of them built on NVIDIA hardware. The AMD-TCS announcement introduces a different architectural approach — open, rack-scale, and built on AMD’s own GPU, CPU, and networking stack — giving enterprises and government-backed AI programmes a credible alternative at meaningful scale.
Helios is powered by AMD Instinct MI455X GPUs, next-generation AMD EPYC Venice CPUs, AMD Pensando Vulcano NICs, and the open ROCm software ecosystem. It is designed as a complete rack-scale platform rather than a collection of components, which means the unit of deployment for enterprises is the full stack rather than individual chips. That distinction matters for data centre build-out timelines and operational consistency, particularly for organisations moving from AI pilots to production workloads at scale.
What HyperVault is and why it exists
TCS established HyperVault in 2025 with a specific mandate: to deliver gigawatt-scale, secure, and reliable AI-ready infrastructure for hyperscalers, AI companies, and global enterprises. The subsidiary is not a rebranding of TCS’s existing data centre capabilities. It is a purpose-built vehicle for the AI infrastructure buildout, designed to operate at a scale and speed that TCS’s main services business was not structured to address.
The Helios deployment at HyperVault will support up to 200 MW of capacity, with both companies working directly with hyperscalers and AI companies to accelerate data centre builds across India. The AI-ready data centre blueprint being developed jointly will be available to those partners, positioning HyperVault as an infrastructure platform rather than simply a facility operator.
K. Krithivasan, MD and CEO of TCS, said: “This collaboration lays the foundation for AMD’s first Helios powered AI infrastructure in India. By combining our strengths in AI, connectivity, sustainable power, and advanced data center engineering, we are poised to deliver infrastructure solutions for AI companies and global enterprises. We are thrilled to deepen our longstanding partnership with AMD as we expand our participation in the AI ecosystem, Infrastructure to Intelligence.”
The open architecture argument
AMD’s pitch for Helios in India rests partly on openness. The ROCm software ecosystem is open-source, and the platform is designed for long-term flexibility rather than lock-in to a proprietary software stack. For enterprises in India building AI infrastructure for the first time, or for government-linked programmes seeking to avoid structural dependency on any single vendor’s ecosystem, that positioning carries practical weight.
The platform is engineered to support the shift from AI experimentation to large-scale deployment, which AMD characterises as the defining infrastructure challenge of the current moment. Organisations that ran successful pilots in controlled environments are now asking how to run the same workloads reliably at ten or fifty times the scale. Helios is positioned as the answer to that specific question.
Dr. Lisa Su, Chair and CEO of AMD, said: “AI adoption is accelerating from pilots to large-scale deployments, and that shift requires a new blueprint for compute infrastructure. With Helios, we are delivering an open, rack-scale AI platform designed for performance, efficiency, and long-term flexibility. Together with TCS, we are enabling enterprises across India to deploy AI at scale today while building the compute foundation of tomorrow.”
What this adds to the India infrastructure picture
India’s AI infrastructure market in early 2026 is being shaped by a handful of very large commitments arriving in close succession. Yotta announced 20,736 NVIDIA Blackwell Ultra GPUs and a $2 billion investment the same week. The G42-Cerebras-C-DAC consortium announced an 8-exaflop sovereign supercomputer days earlier. OpenAI confirmed a data centre partnership with TCS itself for its Stargate initiative.
What the AMD-TCS announcement adds to that picture is architectural diversity. India’s frontier AI infrastructure, until this announcement, was heavily concentrated around a single GPU vendor. The Helios deployment introduces a second credible platform at scale, which matters for long-term resilience and for the competitive dynamics of the market that will develop around this infrastructure over the next several years.
TCS is the connective tissue across several of these announcements, appearing as both a partner to OpenAI through HyperVault for Stargate compute and now as AMD’s primary delivery partner for Helios in India. Whether that dual role creates tension or synergy as these deployments scale is a question the next eighteen months will answer. For now, TCS has positioned itself at the centre of India’s AI infrastructure buildout in a way that no other single organisation has managed.
