Amazon says its own services will run 97% of Amazon's “mission critical databases” by year's end and 88% of its databases previously on Oracle by January
Lauren Feiner / CNBC :
Context & Ripple Effects
This milestone lands midway through a migration Amazon has been telegraphing all year: a January report described [[a:925421|Amazon and Salesforce making significant progress moving off Oracle tech by building their own and using open-source]], and in August sources said Amazon planned to exit Oracle's proprietary database software entirely by Q1 2020. Today's numbers — 97% of mission-critical databases on Amazon's own services by year-end, 88% of the former Oracle fleet by January — turn that plan into a dated, measurable commitment.
First-order effects
- Oracle loses its most demanding reference customer at accelerating pace: within two months, nearly nine-tenths of the databases Amazon once ran on Oracle shift to AWS's own database services.
- Amazon converts an internal cost-and-dependency problem into a sales asset — every migrated workload is a proof point for selling AWS database services to enterprises still on Oracle.
Second-order effects
- The [[a:932820|database management software Amazon built with VMware for companies running on-premises Microsoft, Oracle, and MariaDB databases]] extends the same playbook outward, giving other Oracle customers an Amazon-backed off-ramp rather than waiting for Oracle's cloud.
- Oracle's cloud pivot — the CEO's pledge that 95% of products would be cloud-available — now has to win customers despite its highest-profile former user publicly dismantling the relationship.
Third-order effects
- If the pattern holds, the largest cloud operators become their own infrastructure vendors, shifting the database market from per-license proprietary economics toward self-built and open-source stacks — and handing every large Oracle account a negotiating template backed by a working example.
- The end state, which Amazon's stated Q1 2020 target makes concrete, is a structural decoupling: hyperscale buyers no longer treat legacy database vendors as irreplaceable suppliers but as workloads to be engineered away.
The trend: Hyperscale cloud providers are systematically replacing proprietary licensed databases with self-built and open-source alternatives, then productizing those migrations against the vendors they left.