Analog circuits have been unsharable by construction. Foundry NDAs restrict process details, testbenches behind published results are rarely released, and the sizing knowledge that produced a working schematic lives in the engineer who built it. Analog-DB removes the representation barrier that makes this true.
The design representation stores each circuit as a process-neutral topology with reusable testbenches and a machine-readable datasheet under one schema. A domain-specific language exposes sub-blocks and device sizes as named parameters carrying their matching constraints, making circuits composable and retargetable across process kits without exposing any foundry-confidential values. A schema-governed catalog lets AI agents discover and reuse designs directly. The numbers from the initial release: 68 circuits across 16 classes, 23 circuit-kit bindings on three open PDKs, all meeting their own specification bands at the typical corner. 17 of 23 imported sizings failed their testbenches on import; a gm/ID sizing loop closed all 17 in one to three iterations. A coding agent working from the released artifacts sized op-amp cores for a chopper instrumentation amplifier, found four hand-entry defects in the originals, and identified a missing common-mode feedback loop the sizing-only baseline missed entirely.
Analog IP reuse has been blocked by representation, not by knowledge scarcity. Enough analog circuits exist in published literature to populate a useful open library; the problem has always been that you cannot run them. Analog-DB closes that gap at schematic level and gives the sizing loop a starting point with testbenches attached. Vendors whose business model assumes analog design cannot be systematized are looking at the first credible bootstrapping event for the analog portion of the open-source silicon stack. If the catalog reaches critical mass on open PDKs, the argument for expensive proprietary analog IP libraries collapses faster than anyone in the analog EDA business is currently pricing in.