Advanced Micro Devices Inc. (AMD) is making a big bet on India, and it's bringing an entire rack-scale blueprint to do it. On Monday, the chipmaker announced it's expanding its partnership with Tata Consultancy Services to roll out its newest AI data center design across India, a move squarely aimed at chipping away at Nvidia Corp. (NVDA)'s stranglehold on the AI infrastructure market.
This isn't just about selling more chips. AMD is trying to sell the whole package: a complete system design that customers can deploy at scale without having to figure out all the integration headaches themselves. That's where Helios comes in.
The Full-Stack Play
Helios is AMD's rack-scale architecture, and it's the centerpiece of this India expansion. Think of it as a blueprint for building massive AI data centers that includes everything from GPUs to CPUs to networking gear, all designed to work together seamlessly.
The plan, according to Bloomberg, is to support up to 200 megawatts of AI infrastructure capacity in India. That's not a small number. We're talking about the kind of power and cooling envelope you need for serious training and inference clusters.
The Helios design packs in AMD Instinct MI455X GPUs, next-generation AMD EPYC "Venice" CPUs, and AMD Pensando Vulcano networking. It runs on AMD's ROCm software stack, which the company is positioning as an open alternative to Nvidia's more proprietary ecosystem. The whole thing ties into TCS's HyperVault initiative, which is focused on building AI-ready facilities for hyperscalers and enterprises.
Why This Matters Now
AMD CEO Lisa Su framed it pretty clearly: "AI adoption is accelerating from pilots to large-scale deployments, and that shift requires a new blueprint for compute infrastructure." Translation: companies are done kicking the tires on AI. They're building it out for real, and they need infrastructure that can scale fast.
And here's the thing—AMD is actually gaining ground. Bloomberg noted that Arista Networks Inc. (ANET) has been seeing roughly 20% to 25% of AI chip deployments going to AMD. That's a meaningful shift from 2025, when Nvidia held about 99% of the market. It's still early days, but the momentum is real.
Why India, Why Now
India isn't a random choice. The country ranks third globally in AI competitiveness, trailing only the U.S. and China, according to Stanford University's Institute for Human-Centered AI. More importantly, India has a track record of scaling technology infrastructure quickly, which is exactly what you want when you're trying to build out gigawatts of AI compute capacity.
TCS has been laying the groundwork for this moment through HyperVault, an initiative it established in 2025 with the goal of delivering gigawatt-scale, secure infrastructure for hyperscalers, AI firms, and multinational customers. Late last year, TCS outlined plans to enter the data center market in earnest, targeting as much as 1.2 gigawatts of capacity.
AMD and TCS are positioning the Helios rollout as a way to help both enterprises and public-sector organizations build local compute capacity. The companies say they plan to work with hyperscalers and AI companies to accelerate data center construction in India, using Helios as a repeatable template.
The 200-Megawatt Question
The headline figure—200 megawatts of AI infrastructure capacity—is significant. For context, that's the kind of scale you need to support large training runs and inference workloads that can't afford downtime or bottlenecks. The number shows up in the expanded collaboration details tied to HyperVault's India build-out, and it signals that both companies are thinking beyond proof-of-concept deployments.
This is AMD's broader strategy in action: move beyond selling individual chips and start selling complete solutions that make it easier for customers to deploy AI at scale. If customers can plug in a tested, integrated blueprint instead of piecing together components from different vendors, that lowers the barrier to choosing AMD over Nvidia.
AMD shares were down 1.79% at $203.60 during premarket trading on Tuesday.