How Virgin Media O2 cut compute costs 30% by giving dbt a memory

Virgin Media O2 described how it reduced BigQuery compute costs and build delays across its enterprise dbt environment, which serves more than 300 users, over 100 projects, and 5,500 models supporting 42 million telecom customers. Its previous stateless workflow forced extensive reparsing, recompilation, and reruns for minor model changes, creating redundant builds, high slot consumption, and developer waiting time. By adopting the Rust-based dbt v2 engine and dbt State, which can reuse unaffected work, the company cut a 1,424-model job from about 1 hour 53 minutes to 1 hour 13 minutes while reducing BigQuery slot use from roughly 4,500 to 3,400 slots per hour. Across the migration, it reported approximately 30% lower compute costs, 35% faster runs, 25% less CI/CD dependency, and 35% less developer time spent waiting, although implementation required resolving legacy SQL, technical debt, package compatibility issues, and FinOps integration requirements. The team plans to expand column-level static analysis and explore AI-assisted failure diagnosis, framing the effort as both a cost-saving initiative and a move toward more proactive engineering.

Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.