Extended Q&A on cloud migration and AI
What is the typical cloud migration cost and timeline?
Enterprise cloud migrations typically cost $500K-$10M and take 12-36 months depending on scope. The migration cost components are: (1) assessment and planning (10-15% of budget), (2) application refactoring (40-50%), (3) data migration (15-20%), (4) integration with existing systems (15-20%), (5) training and change management (10-15%). The 6 R's of cloud migration (Rehost, Replatform, Refactor, Repurchase, Retire, Retain) provide a framework for prioritization. Most enterprises follow a 70-20-10 rule: 70% lift-and-shift (Rehost), 20% refactoring for cloud-native benefits, 10% full rewrite (Gartner, 2026).
How should enterprises evaluate AI cloud services?
Key evaluation criteria for AI cloud services: (1) model availability and selection (foundation models, custom models), (2) customization options (fine-tuning, RAG, prompt engineering), (3) data privacy and security (encryption, compliance certifications, data residency), (4) integration with existing data and applications, (5) cost per token / inference, (6) latency and throughput, (7) support and SLAs. Major providers now offer comparable model selection (Claude, GPT, Llama, Mistral) with differentiators in custom hardware (Trainium, TPU, Maia) and enterprise integration (Databricks, Snowflake) (Forrester, 2026).
What is agentic AI?
Agentic AI refers to AI systems that can autonomously plan and execute multi-step tasks using tools and APIs. Examples: customer service agents that can access order databases, software engineering agents that can write and test code, and research agents that can browse the web and synthesize information. Agentic AI requires: (1) foundation model with strong reasoning, (2) tool use / function calling capability, (3) memory and context management, (4) safety guardrails, (5) observability. Major agentic AI platforms: OpenAI Operator, Anthropic Computer Use, Google Gemini Agents, Microsoft Copilot Studio (Gartner, 2026).
| Cloud platform | 2026 AI revenue | Key AI services |
|---|
| AWS | $20B+ | Bedrock, SageMaker, Q for Developers, Trainium |
| Azure | $30B+ (AI business) | Azure OpenAI, Azure AI Foundry, Maia, Copilot |
| Google Cloud | $15B+ | Vertex AI, Gemini, TPU, Agent Builder |
| Oracle OCI | $10B+ | OCI Generative AI, HeatWave, AI Vector Search |
Source: company earnings reports, Q2-Q3 2026.
Cloud cost optimization
- Right-size compute (use recommender tools to identify over-provisioned resources).
- Use committed-use discounts and savings plans for predictable workloads (up to 70% savings).
- Implement auto-scaling and serverless for variable workloads.
- Move cold data to cheaper storage tiers (S3 IA, Glacier, Azure Cool Blob).
- Use spot instances for fault-tolerant batch jobs (up to 90% savings).
- Implement FinOps practices with cost allocation tags and budgets.
- Audit unused resources (orphaned snapshots, idle load balancers, unattached volumes).
- Negotiate enterprise discount agreements (EDAs) for large commitments.
Photo: NASA, Public Domain (NASA), via NASA Image and Video Library (https://images.nasa.gov/details/KSC-2012-3649)