Science & Tech
AMI orders 9,000 NVIDIA Rubin GPUs, plans $8 bn capex
Can India’s AI buildout move from plan to operating capacity?
Can India’s AI buildout move from plan to operating capacity?
AMI is turning a broad AI infrastructure ambition into a concrete bet on compute, power, and scale, with its now anchored by a firm order for 9,000 . The plan matters because the first tranche carries more than $8 billion in , signaling how India’s push for domestic AI capacity is increasingly shifting from announcements to large, capital-intensive buildouts.
The next milestones are delivery of the Rubin systems in early 2027, buildout of the Hyderabad site, and of the first capacity tranche. The reported first facility phase suggests the 200 MW target will be reached in steps rather than all at once. Execution on power, cooling, and integration will determine whether the announced plan becomes usable compute.
The deal was widely read as a milestone for India’s domestic AI compute market because it combined 200 MW of planned capacity, an $8 billion-plus capex plan, and one of the largest reported in Asia. That scale raised the bar for local training and inference capacity and showed how compute, power, and capital are converging in India’s data center market. It also signaled that the country’s AI infrastructure competition is moving into industrial scale.
Coverage identified AM Intelligence as an AI infrastructure platform set up by . That connection matters because it explains where the project’s power and capital backing are coming from, and why the company can frame compute as an extension of energy infrastructure. The story also placed AMI inside the broader Greenko rather than as a standalone startup.
The company said the 9,000 would use , one of the newest frontier compute systems. That choice suggested the Hyderabad facility was designed for large-scale training and heavy , not conventional hosting. The reported Q1 2027 delivery timeline made the order a medium-term supply-chain commitment.
AMI said the first 200 MW tranche would involve more than $8 billion in . The figure underscored how much of the project cost sits in land, power, cooling, data center construction, and together. It also placed AMI among the most ambitious AI infrastructure builders in India.
AMI said the Hyderabad buildout was part of a planned footprint across India, the United States, Finland and Malaysia. It said the near-term target was to bring of capacity to market, making the power requirement a central part of the project’s scale. That figure became the main signal that the company was building industrial AI infrastructure, not a pilot cluster.
The new hardware was tied to AMI’s Hyderabad facility, which the company described as its first . Reporting said the site would be among Asia’s first using the Vera Rubin platform. This anchored the project physically in Hyderabad and turned the strategy into a specific data center footprint.
AMI said it had placed a for 9,000 for its first AI factory in Hyderabad. The company said the systems would be deployed as systems and delivered in the first quarter of 2027. The order marked the project’s shift from planning to a concrete hardware commitment.
The AMI effort was publicly described as part of AM Group, the Greenko-linked energy and infrastructure ecosystem, with a plan to build a renewable-powered AI compute platform. That framing mattered because it tied capital, power, and data center design into one industrial project rather than a simple chip purchase. It also signaled that the company was aiming at , not just internal capacity.
OXMIQ Labs announced a with AM Intelligence Labs, describing itself as the for the compute platform. The announcement said AMI Labs was building toward of AI compute capability by 2030, with an initial 1 GW hub and Phase 1 already in active execution. It also said the platform would combine power generation, data center infrastructure, , software, and .
AM AI Labs, an affiliate of AM Group, received a from the Yamuna Expressway Industrial Development Authority for land to set up the first two phases of its . The reported phases were and , showing the project moving from broad ambition to site planning and capacity design. The same reporting put the broader program at $25 billion and 500,000 high-performance chips.
At Davos, the Uttar Pradesh government said it had signed an with AM Green Group for a in Greater Noida, backed by the founders of Greenko. The plan was framed as a phased build for , with a first tranche expected in 2028 and full scale by 2030. This set the wider strategic backdrop for the family of AI infrastructure bets that later included AMI.