As Snowflake transforms into a full-fledged AI platform, traditional FinOps models, focused on warehouse sizing and compute credits, are no longer sufficient. The rise of Cortex AI introduces token-based economics and multi-dimensional cost drivers, demanding a new FinOps playbook. The workshop guides participants through actionable deliverables for mastering FinOps in the era of Snowflake AI.

1. Understanding the Shift: • Explore the evolution from compute-based to token-based cost models in Snowflake AI. • Identify new cost drivers: token consumption, model complexity, and serving infrastructure.
2. Building AI Cost Observability: • Learn to leverage Snowflake’s ACCOUNT_USAGE telemetry for AI workloads. • Hands-on demonstration: Integrate Cortex AI usage views into FinOps dashboards to monitor token flow and high-cost prompts.
3. Optimizing AI Workloads: • Case study: Analyze a real-world sentiment analysis pipeline to uncover hidden token costs. • Apply optimization strategies: data pre-filtering, text chunking, and model selection to reduce AI spend.
4. Managing Indirect AI Costs: • Address indexing and serving costs in Cortex AI Search. • Implement best practices: minimize warehouse size, optimize refresh frequency, and suspend unused services.
5. Establishing AI-Aware Governance: • Classify AI use cases by business impact and align data governance with AI policies. • Quantify the cost of poor data quality and ensure GenAI readiness.