• How GEICO lowered its $300M cloud spend and decoupled security from the network Mitch Pronschinske Optimize operations Risk & compliance Speed & agility Secrets & identity management Culture & collaboration AI Platform engineering HashiConf Vault Jan 13, 2026 Mitch Pronschinske Share article Twitter share LinkedIn share Facebook share Copy URL GEICO is a juggernaut in the US insurance industry, with an IT infrastructure team that has profoundly transformed its digital architecture over the past decade. • Here’s a snapshot: 2013: 100% on-premises, the CIO decides to go all-in on public cloud 2020: 80% public cloud, distributed across 8 clouds, applications still have to use databases with a client-server model that’s “calling back to the closet of IT in the back room” 2022: Cost structure is 3x higher with 200K cores of compute, 30K+ VMs; they’re spending $300M+ on infrastructure, but developer productivity is down, and they’ve lost two 9s of reliability 2026: Reduced cost to serve, increased portability, improved data quality for eventual AI initiatives, and simplified infrastructure through standardization and centralization GEICO’s cost reduction results and estimations chart. • This is acommon journeyfor enterprises: A lack of standards as you grow the surface area of your cloud for things like network access controls, user authentication, etc. • isn’t uncommon - just take a look at the recentCloud Complexity Report. • “Every datacenter was configured differently. • Every cloud provider was configured differently.

Article Summaries:

  • GEICO is a juggernaut in the US insurance industry, with an IT infrastructure team that has profoundly transformed its digital architecture over the past decade. Here’s a snapshot: - 2013: 100% on-premises, the CIO decides to go all-in on public cloud - 2020: 80% public cloud, distributed across 8 clouds, applications still have to use databases with a client-server model that’s “calling back to the closet of IT in the back room” - 2022: Cost structure is 3x higher with 200K cores of compute, 30K+ VMs; they’re spending $300M+ on infrastructure, but developer productivity is down, and they’ve l

Sources: