Google Cloud

Google Cloud Platform (GCP), launched as App Engine in 2008 and headquartered in Mountain View, California, is the cloud division of Alphabet/Google. Powered by the same infrastructure that runs Google Search, Gmail, YouTube, and Maps, it now holds roughly 13% of global cloud market share - third behind AWS (31%) and Azure (21%) - and generated $58.7 billion in revenue in 2025, its first full year of sustained profitability.

For Indian teams, Google Cloud operates two full-scale regions: asia-south1 (Mumbai) and asia-south2 (Delhi). Both offer Compute Engine, GKE, Cloud SQL, BigQuery, Vertex AI, and Cloud Storage. Mumbai is one of the few regions globally with H100 GPU (A3 instances) and A100 availability. The Delhi region is newer and growing in service coverage. Google has committed $2 billion in India infrastructure investment.

GCP's strongest differentiators are its AI/ML ecosystem - Vertex AI (managed ML platform), Gemini models via Vertex AI Model Garden, TPUs (Tensor Processing Units) for extreme-scale training, and BigQuery for petabyte-scale serverless analytics. It also leads on Kubernetes - Google invented K8s, and GKE (Google Kubernetes Engine) remains the gold standard managed K8s service. Its global private fibre network delivers some of the lowest global latencies of any hyperscaler.

GCP's unique pricing advantages include automatic Sustained Use Discounts (SUDs) - up to 30% off for instances running more than 25% of a month with no commitment - and Spot VMs with up to 91% discount. Committed Use Discounts (CUDs) offer 37-55% savings over 1-3 years. These stack to make GCP highly competitive for long-running steady workloads.

• Two Indian regions - Mumbai (asia-south1) and Delhi (asia-south2) - with H100 GPU available in Mumbai
• Automatic Sustained Use Discounts (SUDs) - up to 30% off with zero commitment for instances running 25%+ of a month
• Spot VMs - up to 91% discount for interruptible workloads; highest spot discount of the three major hyperscalers
• Committed Use Discounts (CUDs) - 37-55% savings over 1-3 year terms; flexible spend-based CUDs across all eligible compute
• Vertex AI - fully managed ML platform for training, tuning, evaluation, and deployment; integrates with Gemini, Llama, and open models
• TPUs (Ironwood, v5e, v5p) - Google's own AI silicon delivering the best price-performance for large-scale transformer training
• BigQuery - the industry-leading serverless analytics engine; query petabytes with zero infrastructure management; pay per query
• GKE (Google Kubernetes Engine) - the gold standard managed K8s; Google invented Kubernetes; deepest K8s expertise of any cloud
• Cloud Run - fully managed serverless containers; GPU support in Mumbai and Delhi regions now available by invitation
• Custom machine types - configure exact vCPU and RAM combinations to avoid paying for resources you don't need
• Global private fibre network - Google's own undersea cables and global backbone deliver some of the lowest inter-region latencies
• $300 free trial credit for 90 days + Always Free tier - e2-micro VM always free in select regions; most generous credit of the three hyperscalers
• Anthos - manage workloads across GCP, AWS, Azure, and on-premises from a single control plane; unique multi-cloud capability

• USD billing - no INR pricing, no UPI/NEFT, no GST-compliant invoices; same forex risk as AWS and Azure
• Not MeitY empanelled - ineligible for Indian government and PSU procurement without separate assessment
• No Indian data sovereignty guarantee - US-headquartered; DPDP Act compliance requires careful assessment, DPAs, and contractual commitments
• H100 GPU in Mumbai is invitation-only for Cloud Run - some GPU features require contacting the Google account team; not fully self-serve
• Egress charges - outbound data transfer to the internet starts at $0.12/GB for the first 1TB; intra-region GCP traffic is free
• SUDs and CUDs cannot be combined - you must choose one discount model; CUDs are usually better for full-time workloads but require upfront commitment
• Smaller service catalogue than AWS - GCP has ~150 services vs AWS's 200+; some enterprise niche services only exist on AWS
• Support costs - basic support is free but limited; Enhanced ($150/mo) or Premium (custom) support needed for production SLAs and faster response

All prices in USD per month, on-demand. Mumbai region typically adds ~15-20% premium over us-central1. Per-second billing (1-minute minimum). Persistent disk storage billed separately ($0.08–0.17/GB-month). Auto-applied SUDs where eligible.

E2 - General Purpose (Lowest cost, best for dev/web/apps) · Mumbai On-Demand

* e2-micro always free in us-west1, us-central1, us-east1 only - not Mumbai. Mumbai e2-micro billed at standard rate.

N2 - General Purpose (Intel, SUD-eligible) · Mumbai On-Demand

GPU & Accelerator Instances - On-Demand

Key Managed Services - Starting Prices

* Cloud Run GPU in Mumbai and Delhi is currently invitation-only - contact your Google account team. A3 Mega (H100) and A2 Ultra (A100) in Mumbai available as standard Compute Engine instances. Full list at cloud.google.com/about/locations.