Two things landed in the last nine months that change how anyone running a data center hall has to think about measurement and verification (M&V). First, the U.S.
Every "AI for buildings" pitch you got this quarter quietly assumes something that most portfolios do not have: a clean, vendor-neutral data layer the model can actually read.
In 2026 the building digital twin stops being a dashboard you look at and starts being a controller that acts. The proof is showing up first where the economics are brutal and the physics are unforgiving — AI data center cooling — and the playbook…
The 2026 sensor-fusion conversation has been hijacked by accuracy benchmarks — "we hit 99%!" — when the number that actually decides whether your project ships on time is the privacy class of the streams you fuse.
Measurement & verification is quietly splitting into two worlds. The first is the old world of manual IPMVP Option C studies — engineers pulling utility bills, building regression models in a spreadsheet, and waiting months for a savings number.
Every AI-HVAC vendor pitch in 2026 cites "verified savings" and "IPMVP-aligned M&V." Almost none cite the two numbers that decide whether the verification actually held: CV(RMSE) and NMBE .
The 2026 occupancy-intelligence stack has converged on a multi-source default: sensors + badge + WiFi + booking, all in one dashboard. The trouble is they routinely produce three different numbers for the same Tuesday at 2pm.
For two decades, IPMVP Option D — calibrated simulation — was the protocol no banker would accept. Too many knobs, too much modeller discretion, no clean counterfactual.
The sensor-fusion conversation in 2026 stopped being academic. Three concrete vendor releases — VergeSense's native Juniper Mist WiFi ingest (Feb 2026), MultiTech's LoRaWAN-into-Niagara driver (Apr 2026), and UL-recognized A2L refrigerant leak…