Systems in the built environment continuously emit time series data about resource usage (e.g., energy and water), embedded electrical generation/storage, status of equipment, patterns of building occupancy, and readings from IoT sensors. This presents opportunities for new analytics and supervisory control applications that help reduce greenhouse gas emissions due to energy demand, if the barrier of data heterogeneity can be overcome. Semantic models of buildings -- representing structure, integrated equipment, and the many internal connections -- can help achieve interoperable data re-use by describing overall context, in addition to metadata. In this paper, we describe the Data Clearing House (DCH), a semantic building platform that hosts sensor data, building models, and analytics applications. This fulfills the key phases in the lifecycle of semantic building data, which includes: cost-effective ingestion of Building Management System (BMS), IoT, metering and meteorological time series data from a wide range of open and proprietary systems; importing and validating semantic models of sites and buildings using the Brick Schema; interacting with a discovery API via a high-level domain-specific query language; and deploying applications to modelled buildings. Having onboarded multiple buildings belonging to our own organisation and external partners, we are able to comment on the challenges to success of this approach. As an example use-case of the semantic building platform, we describe a measurement and verification (M&V) application implementing the 'whole facility' (Option C) method of the International Performance Measurement and Verification Protocol (IPMVP) for evaluating electrical metering data. This compares energy consumption between nominated baseline and analysis time periods, to quantify the energy savings achieved after implementing an intervention on a site.