Cloud Provider · 3 of 6

Google Cloud (GCP)

The data & AI cloud. Smaller than AWS or Azure, but home to BigQuery, the most-loved warehouse in tech, plus first-party Gemini models and a deeply Kubernetes-native compute story.

#3 by shareBigQuery / DataVertex AI / GeminiKubernetes-nativeDeveloper-friendly
← Back to Cloud
Quick Facts

At a Glance

Launched
2008 (App Engine)
Parent
Alphabet / Google
Regions
~40+ globally
Market share
~11%
Strengths
Data, AI, Kubernetes
Native AI
Gemini family + Vertex AI

Basic Concepts

  • Organization → Folder → Project → Resources — every resource lives inside a Project.
  • IAM assigns roles to identities at any level — inherited downward.
  • Service accounts for service-to-service auth; gcloud CLI for everything.
  • Kubernetes was born here — GKE is widely considered the best managed K8s.
  • Network is global — same VPC can span regions; one global load balancer per app.
Services

Signature Services

CategoryServiceUse
ComputeCompute EngineVMs.
ComputeCloud RunServerless containers — GCP's flagship compute.
ComputeCloud Functions / Cloud Run FunctionsFaaS.
ContainersGKE (Standard & Autopilot)Managed Kubernetes.
StorageCloud StorageObject storage with strong consistency.
DatabaseCloud SQL / AlloyDBManaged Postgres / MySQL / SQL Server.
DatabaseSpannerGlobally distributed, strongly consistent SQL.
DatabaseFirestore / BigtableServerless document / wide-column.
AnalyticsBigQueryServerless data warehouse — GCP's killer app.
AnalyticsDataflow / Dataproc / Pub/SubBeam streaming, Spark, messaging.
AI / MLVertex AIGemini, hosted models, training, deployment.
NetworkingVPC, Cloud Load Balancing, Cloud CDN, Cloud ArmorNetwork, global LB, CDN, WAF.
IdentityCloud IAM, Identity PlatformService / app identity.
DevOpsCloud Build, Artifact Registry, Cloud DeployCI/CD & image registry.
Mechanics

The GCP Way

BigQuery — Worth Its Own Section

Probably the single biggest reason teams pick GCP. Pure SQL, no servers, scales to petabytes:

  • Pay per scanned bytes (or flat-rate slot reservations).
  • Federated queries across Cloud Storage, Bigtable, Spanner, even external clouds (Omni).
  • BigQuery ML lets you train models with SQL.
  • Streaming inserts for near-real-time data.
  • Best-in-class for analyst-friendly cloud data work.
Cloud Run — The Modern Default

Bring a container, get a URL. Auto-scales 0 → N, billed per request, supports gRPC, websockets, jobs. For most teams it replaces both App Engine and Cloud Functions.

Networking
  • VPC is global, not regional like AWS — simpler multi-region design.
  • Global HTTP(S) Load Balancer = one anycast IP, routes worldwide.
  • Private Service Connect for private connectivity to managed services.
  • VPC Service Controls create a perimeter around services to prevent data exfiltration.
Vertex AI & Gemini

The unified AI platform — Gemini model API, model garden (Claude, Llama, Mistral, etc.), training, fine-tuning, evaluation, vector search, agents builder. Tightly integrated with BigQuery for "RAG over my warehouse."

Pricing Notes
  • Sustained-use discounts — auto-applied for VMs that stay on.
  • Committed-use discounts — explicit 1- or 3-year commit.
  • Spot VMs & Spot GKE nodes — big savings for batch.
  • BigQuery — slot reservations beat on-demand for steady workloads.
  • Egress still expensive (better than AWS in many lanes, but still real).
When to Pick

Where GCP Wins

Data & Analytics

BigQuery is the single best reason to be on GCP.

AI & ML

Vertex AI + Gemini + GPUs / TPUs.

Kubernetes-Heavy

GKE Autopilot is the smoothest managed K8s.

Container-First Apps

Cloud Run is a developer-experience standout.

Global Apps

One global VPC + one global LB = simpler architecture.

Open-Source-Friendly

Most CNCF-aligned cloud (K8s, Istio, Knative, gRPC, Beam, TensorFlow…).

Continue

Other Cloud Providers

AWSMicrosoft AzureOracle CloudDigitalOcean / LinodeVercel / Netlify / Cloudflare ↑ Back to Cloud