What’s new

In back-to-back announcements, two U.S. hyperscalers made their biggest India commitments yet:

  • On December 9, 2025, Microsoft said it will invest $17.5 billion over 2026–2029 to expand AI-ready cloud infrastructure, launch sovereign cloud options, and double its India AI-skilling goal to 20 million people by 2030. Reuters | Microsoft
  • On December 10, 2025, Amazon committed “more than $35 billion” across its India businesses through 2030, targeting AI-driven digitization, logistics, exports, and jobs. Reuters | About Amazon | About Amazon India
A stylized India map glowing with data center clusters and fiber links, symbolizing hyperscaler investment.
20.47B
UPI transactions (Nov 2025)Source: NPCI-via-LiveMint-2025-12-02

Why India is the prize for AI and cloud

India combines massive digital demand with maturing policy and infrastructure:

  • A rapidly expanding online population will exceed 900 million users in 2025, with rural India now over half of the base. Business Standard
  • UPI has become the world’s most active real-time payments system and crossed 20.47 billion transactions in November 2025 alone. LiveMint
  • India has operationalized the Digital Personal Data Protection (DPDP) Rules, 2025 with a phased compliance roadmap, clarifying privacy, breach reporting and governance obligations for platforms. Business Standard | Reuters
  • Data-center capacity is sprinting toward roughly 1.8 GW by 2027, powered by cloud and AI workloads. JLL
~1.8 GW
India DC capacity by 2027Source: JLL-India-DC-2025

Inside the bets: who’s building what

Microsoft: $17.5B over four years (2026–2029)

  • Infra: A new Azure region in Hyderabad (India South Central) goes live mid-2026, alongside expansions in existing India regions. Microsoft also introduced Sovereign Public Cloud and Sovereign Private Cloud to address compliance and data-residency needs. Microsoft | Azure regions background
  • Skills: Doubling its AI-skilling goal to 20 million Indians by 2030. Reuters
  • Public sector: Collaboration to infuse AI into Ministry of Labour platforms (e-Shram and National Career Service) aimed at serving over 310 million informal workers. Microsoft

Amazon: >$35B through 2030 (across all India businesses)

  • Scope: Investments span e-commerce, logistics, payments, content, and AWS—centered on AI-enabled seller tools, customer experiences, and export acceleration. About Amazon | Reuters
  • Economic goals: Lift cumulative e-commerce exports enabled to $80B, take AI benefits to 15 million small businesses, and support up to 3.8 million direct and indirect jobs by 2030. About Amazon India
  • Cloud baseline: AWS separately plans $12.7B in India cloud infrastructure investment by 2030 (announced 2023), signaling ongoing capacity growth for developers and enterprises. AWS India press
  • Energy: Amazon has built a 1+ GW portfolio of wind and solar projects in India and continues to add new capacity to match its electricity use with renewables. Amazon sustainability update | About Amazon India

Also in the mix: Google’s $15B AI hub

Google announced a $15B AI hub in Visakhapatnam, including gigawatt-scale compute and a new international subsea gateway to diversify India’s connectivity beyond Mumbai and Chennai. Google blog | AP/ABC News

The big India bets at a glance

CompanyHeadline commitmentTimelineFocusNotable details
Microsoft$17.5B2026–2029AI infra, sovereign cloud, skillsNew Hyderabad region (mid-2026), 20M people to be AI-skilled by 2030. Microsoft
Amazon>$35B (all India businesses)Through 2030AI-driven commerce, logistics, exports, jobsExports target $80B; AI tools for 15M SMBs; up to 3.8M jobs supported. About Amazon India
AWS (subset of Amazon)$12.7BThrough 2030Cloud regions, capacityBuilds on Mumbai & Hyderabad regions; training programs and renewable PPAs. AWS India
Google$15B2026–2030AI hub & subsea gatewayVizag campus with 1-GW-class capacity and new cable landing. Google

What this means for automation and productivity

  • Lower-latency AI and more GPU access: New regions and capacity should reduce model-inference latency for Indian and nearby users, while easing some GPU scarcity for training and fine-tuning.
  • “Sovereign-by-design” enterprise stacks: Microsoft’s sovereign cloud options, plus India’s DPDP framework, make it easier for regulated sectors (BFSI, healthcare, public agencies) to adopt AI safely.
  • SMB digitization at scale: Amazon’s targets (AI tools for 15 million small businesses; exports to $80B) point to a wave of AI-assisted listing, pricing, customer support, and logistics automation for India’s long tail of sellers.
  • Talent flywheel: Government-backed IndiaAI Mission and corporate skilling programs expand the pool of AI engineers, MLOps specialists, and prompt/application designers.
TipFor CIOs and operations leaders
  • Negotiate capacity reservations for bursty AI workloads near Mumbai/Hyderabad/Vizag.
  • Adopt data-classification and residency controls now to align with DPDP rules phasing in through 2026–2027.
  • Pilot sovereign patterns (key management, logging, isolation) where regulatory interpretation is evolving.
  • Benchmark multi-cloud GPU pricing (H100/H200/MI300X/TPUs) and design portability into your AI stack.
  • Tie green-PPA or renewable matching to your India hosting strategy to derisk energy constraints and ESG commitments.

Power, policy, and practical constraints

  • Grid and land: India added a record 31+ GW of non-fossil capacity this year, but hyperscale AI clusters still face transmission and siting constraints in specific corridors. Expect PPAs, on-site generation, and location diversity. Times of India | Reuters
  • Privacy and governance: The DPDP Rules, 2025 are in force with staged compliance windows (some obligations extend into 2027). Plan for consent, breach reporting, children’s data, and Board oversight. Business Standard
  • Chip supply: India remains compute-constrained relative to demand. Local access improves with new regions, but advanced AI-training capacity will still be globally orchestrated. Reuters

A pragmatic adoption playbook

  1. Map latency-sensitive services to the nearest region (Mumbai/Pune/Chennai now; Hyderabad mid-2026; Vizag as Google’s hub comes online) and use traffic shaping to keep user P95 low.
  2. Classify data under DPDP and sectoral norms; align key management (KMS/HSM), audit, and DLP to sovereign patterns where required.
  3. Build a portable AI stack (containers or serverless + vector DB + retrieval) and validate failover across at least two providers/regions.
  4. Tie compute forecasts to product roadmaps; reserve capacity for peaks (sales, festivals) and negotiate committed-use discounts.
  5. Invest in skills: prioritize MLOps, governance, and prompt-engineering; Microsoft’s and Amazon’s India programs can offset training costs and hiring gaps.

Sources