Azure vs GCP
Microsoft's enterprise cloud versus Google's engineering cloud. One sells to CTOs, the other sells to engineers. Guess which one makes more money.
GCP
GCP wins for technical teams building modern applications. Better Kubernetes (they invented it), better AI/ML platform (Vertex AI, TPUs), cleaner developer experience, and more honest pricing. Azure wins the enterprise sales game because Microsoft bundles it with Office 365 and has 10x the sales force. If your CTO already signed an Enterprise Agreement, this comparison is moot — you are on Azure.
The Honest Framing
Azure has 24% cloud market share. GCP has 11%. AWS has 31%. Azure is winning the enterprise race because Microsoft has a sales army and an Office 365 bundle that makes Azure credits essentially free for big companies.
But market share is not the same as technical merit. Plenty of terrible products have dominant market share (looking at you, Oracle).
If someone else is choosing your cloud: it's Azure, because your company already has a Microsoft Enterprise Agreement. If YOU are choosing your cloud: read on.
Kubernetes: GCP Wins, Obviously
Google invented Kubernetes. GKE (Google Kubernetes Engine) is the gold standard. It was the first managed Kubernetes service, it auto-upgrades, it has Autopilot mode (fully managed nodes), and it just works.
AKS (Azure Kubernetes Service) has improved dramatically but still has more rough edges. Node pool management is clunkier, networking is more complex (Azure CNI vs kubenet decisions), and the Azure CLI for AKS operations is verbose.
GKE Autopilot: you do not manage nodes at all. You submit pods, Google figures out the rest. AKS has no equivalent. EKS (AWS) has no equivalent. This alone is worth choosing GCP for Kubernetes workloads.
AI/ML: GCP Is Playing a Different Game
GCP has TPUs. Nobody else does. For training large models, TPU v5e pods are significantly cheaper than GPU clusters on Azure or AWS.
Vertex AI is a comprehensive ML platform with AutoML, custom training, model serving, feature store, and ML pipelines. Azure ML is comparable in features but the UX is worse (everything in Azure has worse UX).
BigQuery ML lets you run ML models with SQL. Not great ML, but great for analysts who know SQL and want predictions without learning Python.
Azure's advantage: OpenAI partnership. Azure OpenAI Service gives you GPT-4o, DALL-E, and Whisper with enterprise SLAs. If you need OpenAI models with compliance guarantees, Azure is the only option.
What Nobody Tells You
Azure's portal is a disaster. It is slow, inconsistent, and requires clicking through 47 blade panels to accomplish anything. The Azure CLI is verbose (az group create --name mygroup --location eastus vs gcloud compute instances create myvm). PowerShell integration is good if you are already a PowerShell shop; infuriating if you are not.
GCP's console is cleaner and faster. But GCP has its own problem: product shutdowns. Google kills products. They killed Cloud IoT Core, Stadia, and dozens of other services. Every time you adopt a GCP service, you wonder if it will exist in 3 years.
Azure will never kill a product because enterprise customers would riot. This matters for long-term planning.
Also: GCP's support is worse than Azure's. Azure support with a Premier Agreement gets you a dedicated TAM. GCP support feels like yelling into a void unless you are on the Premium tier ($12,500/mo minimum).
Data and Analytics
BigQuery is GCP's killer feature. Serverless, petabyte-scale, SQL-based analytics. You upload data, you query it. No cluster management, no capacity planning. Pay per query.
Azure Synapse Analytics is the equivalent and it is fine. But it requires more configuration, more understanding of distribution strategies, and more operational overhead.
For data warehousing: BigQuery > Synapse. For data lakes: both are comparable (GCS + Dataproc vs ADLS + HDInsight/Databricks).
Azure's advantage here is Power BI integration. If your org uses Power BI (and most enterprises do), Azure's data stack connects natively. GCP + Looker works but it's a newer integration.
Pricing: Both Are Confusing, But Differently
Azure pricing is intentionally opaque. There are 14 VM series, each with sub-variants, spot pricing, reserved instances, savings plans, and hybrid benefit discounts. The pricing calculator requires a PhD.
GCP pricing is simpler on the surface — sustained use discounts are automatic (no commitment needed, you just get cheaper the longer a VM runs). Committed use discounts require 1-3 year commitments.
Real-world costs: GCP is typically 10-20% cheaper for compute. Azure is cheaper if you have Enterprise Agreement credits. Both are more expensive than AWS for equivalent workloads.
Free tiers: GCP gives you $300 in credits for 90 days plus an always-free tier (e2-micro VM, 5GB Cloud Storage). Azure gives you $200 for 30 days plus a free tier. GCP's free tier is more generous.
Egress: Both charge for data leaving their cloud. GCP charges $0.12/GB. Azure charges $0.087/GB. Both are highway robbery, but Azure is slightly less criminal.
Quick Comparison
| Factor | Azure | GCP |
|---|---|---|
| Kubernetes | AKS (good) | GKE (best-in-class) |
| AI/ML Platform | Azure ML + OpenAI | Vertex AI + TPUs |
| Enterprise Sales | Dominant (EA bundles) | Weaker sales org |
| Data Warehouse | Synapse (complex) | BigQuery (elegant) |
| Console/UX | Slow, cluttered | Clean, fast |
| Product Stability | Never kills products | Google kills things |
| Support Quality | Good (with Premier) | Weak (unless Premium) |
| Free Tier | $200/30 days | $300/90 days |
The Verdict
Use Azure if: Your company has a Microsoft Enterprise Agreement, you need OpenAI models with enterprise SLAs, your org lives in the Microsoft ecosystem (AD, Office 365, Power BI), or you need guaranteed long-term product stability.
Use GCP if: You are a technical team choosing on merit. You need the best Kubernetes experience, serious AI/ML infrastructure (TPUs), or BigQuery for analytics. You value developer experience over enterprise sales relationships.
Consider: AWS is still the safe default with the largest ecosystem. If none of the Azure/GCP-specific advantages matter to you, AWS is the boring correct choice.
GCP wins for technical teams building modern applications. Better Kubernetes (they invented it), better AI/ML platform (Vertex AI, TPUs), cleaner developer experience, and more honest pricing. Azure wins the enterprise sales game because Microsoft bundles it with Office 365 and has 10x the sales force. If your CTO already signed an Enterprise Agreement, this comparison is moot — you are on Azure.
Related Comparisons
Disagree? nice@nicepick.dev