Google Cloud Professional Cloud Architect (PCA) is Google Cloud's flagship design certification. It tests whether you can turn business and technical requirements into a secure, reliable and cost-effective Google Cloud architecture, plan migrations, and guide teams through implementation and operations. The current exam guide adds a lot of generative AI: Gemini models, Model Garden, the Gemini Enterprise Agent Platform, AI Hypercomputer and securing AI with Model Armor.
This guide follows Google's official exam guide: all six sections with weights, the four case studies, the design decisions to master, an 8-week plan and 15 original scenario questions.
Quick facts (Google Cloud, November 2026)
| Item | Detail |
|---|---|
| Length | 2 hours |
| Questions | 50–60, multiple choice and multiple select |
| Case studies | 2 per exam, making up 20–30% of the questions, shown on a split screen |
| Fee | USD 200 (plus tax where applicable) |
| Languages | English, Japanese |
| Delivery | Online-proctored or at a test centre |
| Prerequisites | None; 3+ years of industry experience including 1+ year designing and managing Google Cloud solutions recommended |
| Valid for | 2 years |
| Renewal | Standard exam, or a renewal exam (1 hour, 25 questions, USD 100, built around a generative AI case study) from 60 days before expiry |
See the official Professional Cloud Architect page and its exam guide before booking.
Who should take PCA?
- Cloud engineers and architects who design solutions on Google Cloud.
- AWS or Azure architects adding a second cloud (see which cloud to learn first).
- People who have passed the Associate Cloud Engineer and have a year or more of real project work.
Familiarity with the Google Cloud Well-Architected Framework is described as a key requirement. Its pillars (operational excellence, security, reliability, performance optimisation, cost optimisation and sustainability) run through every section.
Sections and weights
| Section | Weight |
|---|---|
| 1. Designing and planning a cloud solution architecture | ~25% |
| 2. Managing and provisioning a cloud solution infrastructure | ~17.5% |
| 3. Designing for security and compliance | ~17.5% |
| 4. Analyzing and optimizing technical and business processes | ~15% |
| 5. Managing implementation | ~12.5% |
| 6. Ensuring solution and operations excellence | ~12.5% |
The four case studies
Each exam uses two of these four fictitious companies. Google says several of them use generative AI to solve their problems. Read all four in the official exam guide before the exam, so you can scan them quickly on test day:
- Altostrat Media
- Cymbal Retail
- EHR Healthcare
- KnightMotives Automotive
For each one, write down the business goals, technical requirements, compliance constraints and existing systems. Most case-study questions ask which option best meets a stated requirement, so the winning answer is usually the managed service that satisfies the constraint with the least operational work.
Section 1: Designing and planning (~25%)
- Business requirements: use cases, functional and non-functional requirements, business continuity, cost, integration, data movement, trade-offs, build/buy/modify/deprecate decisions, KPIs and ROI, security and observability.
- Technical requirements: Well-Architected Framework, high availability and failover, elasticity, scalability, latency, backup and recovery, and Gemini Cloud Assist.
- Network, storage and compute: hybrid and multicloud integration; Gemini models, Agent Builder, Model Garden and AI Hypercomputer; VPC, peering, Shared VPC, Private Service Connect, firewalls and load balancers; storage choices; GKE vs Cloud Run vs Cloud Run functions; Spot VMs and custom machine types.
- Migration plans: Migration Center, testing, network and dependency planning, licensing and financial impact.
- Future improvements: cloud-first design and evolving business needs.
| Need | Usual first choice |
|---|---|
| Stateless container, scale to zero | Cloud Run |
| Complex microservices, custom networking, GPUs | GKE |
| Lift-and-shift VM, licensed software | Compute Engine |
| Global relational database, strong consistency | Spanner |
| Regional PostgreSQL or MySQL | Cloud SQL (or AlloyDB for heavier PostgreSQL) |
| Analytics warehouse | BigQuery |
| Wide-column, very high throughput time series | Bigtable |
| Object storage and data lake | Cloud Storage |
Section 2: Managing and provisioning (~17.5%)
- Networks: hybrid connectivity to on-premises, multicloud links, firewalls and intrusion protection, VPC design and load balancing.
- Storage: allocation, access, transfer and latency, retention and lifecycle, growth and backup.
- Compute: Spot vs standard, networking for Compute Engine, GKE, serverless and VMware Engine, patching, container orchestration and serverless.
- ML workflows on the Gemini Enterprise Agent Platform: Agent Platform Pipelines, data integration, GPUs and TPUs through AI Hypercomputer.
- Prebuilt AI: Google AI APIs (Search, Conversation, Vision, Image, Video, Audio), Gemini Enterprise features such as AI agents and NotebookLM, and Model Garden models.
Section 3: Security and compliance (~17.5%)
- Security: IAM and the resource hierarchy; key and secret management; separation of duties; VPC Service Controls, organisation policies and hierarchical firewall policies; CMEK with Cloud KMS; Identity-Aware Proxy, service account impersonation and Workload Identity Federation; software supply chain; securing AI with Model Armor and Sensitive Data Protection.
- Compliance: health and children's privacy, data privacy and sovereignty, card data and PII, SOC 2, and audit logs.
For Indian projects, data-sovereignty questions often map to keeping data in asia-south1 (Mumbai) or asia-south2 (Delhi) with an organisation policy that restricts resource locations. See also the DPDP Act for AI applications.
Section 4: Technical and business processes (~15%)
- Technical: SDLC, CI/CD, root cause analysis, testing and validation, service catalogue and disaster recovery.
- Business: stakeholder and change management, team skills, decision-making, customer success, CapEx vs OpEx and business continuity.
Section 5: Managing implementation (~12.5%)
- Advising teams: deployment, Apigee API management, load, unit and integration testing, migration tooling and Gemini Cloud Assist.
- Working programmatically: Cloud Shell and Cloud Code, the
gcloud,gsutilandbqtools, emulators for Bigtable, Spanner, Pub/Sub and Firestore, Terraform, and Google API client libraries.
Section 6: Solution and operations excellence (~12.5%)
- The operational excellence pillar of the Well-Architected Framework.
- Google Cloud Observability: monitoring, logging, profiling, benchmarking and alerting.
- Deployment and release management, supporting deployed solutions and quality control.
- Reliability in production: chaos engineering, penetration testing and load testing.
8-week study plan
| Week | Focus | Exercise |
|---|---|---|
| 1 | Well-Architected Framework, resource hierarchy, IAM | Design folders, projects and groups for a company with dev, test and prod |
| 2 | Compute choices: Compute Engine, GKE, Cloud Run, functions | Deploy the same app on Cloud Run and GKE; compare cost and effort |
| 3 | Storage and databases | Decision table for 10 workloads (Cloud SQL, Spanner, BigQuery, Bigtable, Firestore) |
| 4 | Networking: Shared VPC, PSC, load balancers, hybrid connectivity | Draw a hub-and-spoke design with on-premises connectivity |
| 5 | Security: VPC Service Controls, KMS, IAP, Workload Identity Federation, organisation policies | Lock a BigQuery dataset inside a service perimeter |
| 6 | Generative AI: Gemini models, Model Garden, Agent Platform, Model Armor | Architecture for a grounded support agent with safety controls |
| 7 | Migration, DR, observability, release management | Migration plan and DR design for a three-tier app |
| 8 | All four case studies, sample questions | Answer every case study's requirements on one page each; book the exam |
15 scenario questions with answers
These are original scenarios written for this guide, not exam questions.
1. A stateless API has unpredictable traffic and should cost nothing when idle. Which compute option?
Cloud Run.
2. A bank needs a relational database with strong consistency across several Regions. What do you recommend?
Spanner (multi-region configuration).
3. Several projects must share one network managed by a central team. What do you use?
Shared VPC, with a host project and service projects.
4. Data in BigQuery must not be copied out to projects outside the organisation, even by users with access. What prevents this?
VPC Service Controls service perimeters.
5. A CI system on another cloud must deploy to Google Cloud without service account keys. How?
Workload Identity Federation.
6. Employees need secure access to an internal web app without a VPN. What do you use?
Identity-Aware Proxy (with context-aware access if needed).
7. Regulators require that encryption keys are controlled and rotatable by the customer. What do you use?
Customer-managed encryption keys (CMEK) in Cloud KMS.
8. Resources must only be created in Indian Regions. How do you enforce it?
An organisation policy restricting resource locations to asia-south1 and asia-south2.
9. A batch job can tolerate interruption and must be as cheap as possible. Which VM type?
Spot VMs.
10. A team needs to call a third-party open model and Gemini from one place with consistent governance. Where do they find and deploy models?
Model Garden on the Gemini Enterprise Agent Platform.
11. A customer-facing AI agent must be protected from prompt injection and must not leak PII. What do you add?
Model Armor for prompt and response screening, plus Sensitive Data Protection for PII.
12. A company wants to inventory on-premises servers and estimate migration cost. What do you use?
Google Cloud Migration Center.
13. Many internal teams consume the same APIs and need keys, quotas and analytics. What do you recommend?
Apigee API management.
14. Developers want to test Pub/Sub and Firestore code locally without cloud costs. What do they use?
The Cloud emulators for Pub/Sub and Firestore.
15. Before launch you must prove the system survives the loss of a zone. What practice applies?
Chaos engineering and failure testing (for example, deliberately removing a zone's instances) with load testing.
After PCA
- Strengthen GCP interviews with Google Cloud interview questions and GCP AI interview questions.
- Read GCP for AI engineers and EKS vs AKS vs GKE.
- See live cloud jobs.
Cloud Soft Solutions in Ameerpet, Hyderabad teaches Google Cloud alongside AWS and Azure in its multi-cloud programs. See all courses and the APEX program.
📞 Call +91 96660 19191 · WhatsApp · Flat No. 513, 5th Floor, Nilagiri Block, Aditya Enclave, Ameerpet, Hyderabad
Frequently Asked Questions
How long is the Professional Cloud Architect exam?
2 hours, with 50 to 60 multiple-choice and multiple-select questions.
How much does the Google Cloud Architect exam cost?
USD 200, plus tax where applicable.
Which case studies are used in the PCA exam?
Altostrat Media, Cymbal Retail, EHR Healthcare and KnightMotives Automotive. Each exam uses two of them, making up 20–30% of the questions.
Which section has the most weight?
Designing and planning a cloud solution architecture, at about 25%.
How long is the PCA certification valid?
2 years. You can renew with the standard exam or a shorter 1-hour renewal exam from 60 days before expiry.
Is there a prerequisite for Professional Cloud Architect?
No. Google recommends 3+ years of industry experience, including 1+ year designing and managing Google Cloud solutions.
