Researched by BEAST Library Curator | Verified by Harper | Quality: 8.6/10
The Execution Gap: Why Building-as-a-Service Vendors Are Buying Their Way Out of the $11 Billion Problem
BLUF: The commercial building industry does not have a diagnosis problem anymore. Fault-detection and analytics platforms — SkyFoundry, Clockworks, Gridium, Carbon Lighthouse — have gotten good at finding the savings. The problem is that almost none of it gets fixed. A 2025 VTS survey of 400 property managers found 61% of their time still goes to work-order management — not to acting on the $11 billion in identified, achievable annual savings sitting in U.S. commercial buildings. The Building-as-a-Service (BaaS) category's answer, visible in two acquisitions inside the last 18 months, is to stop selling diagnosis as a standalone product and start owning the dispatch-and-close-out workflow that turns a flagged fault into a completed work order. If your building's platform still ends at a dashboard, you are paying for half the product.
The gap is not technical — it's operational
The diagnostic layer of the smart-building stack has matured faster than the execution layer underneath it. Fault-detection and diagnostics (FDD) tools reliably surface the same finding across thousands of buildings: a stuck damper, a simultaneous heating-cooling fight, a scheduling error running HVAC through an empty floor overnight. None of that is new news to an experienced facility manager (FM). What has changed is the sheer volume of findings a modern FDD platform can generate — and the analysis in a widely-cited 2026 proptech thesis puts a number on the resulting bottleneck: with 61% of property-manager time consumed by work orders and invoice processing (per the VTS 2025 Global Workplace Report), and the average property manager juggling 5.22 different systems daily, the identified $11B in annual savings sits in a queue, not a completed project.
This is the practitioner's version of a well-known SaaS failure mode: a tool that generates insight faster than the organization can consume it just becomes another alert to triage. The FDD platforms did their job. The industry built the wrong next layer.
What the M&A pattern tells you about where the value is moving
Two acquisitions bracket the shift. Trane Technologies acquired BrainBox AI in January 2025, folding an autonomous-optimization engine into a company that already owns chillers, controls, and — critically — the service technicians who show up on-site. Fifteen months later, Johnson Controls acquired Nantum AI (announced April 27, 2026), pulling the New York-based optimization platform — originally built in-house at Rudin Management starting in 2016 — into the OpenBlue ecosystem. Early pilots reported are modest and specific: 10-15% energy savings in hospital and office pilots while holding comfort steady, with commercial rollout beginning Q3 2026.
Neither deal was bought for the algorithm alone. Both were bought because the acquirer already controls the piece that turns an optimization recommendation into a physical action: technicians, service contracts, and — this is the part FMs should watch closest — the BMS install base the recommendation has to write back into. An independent optimization layer with no execution arm is now competing against incumbents who can dispatch the fix the same afternoon the fault is detected.
The independents are answering with the same move, from the other direction
Not every credible player is getting acquired. Noda, working with Lawrence Berkeley National Lab, published a field study showing a 40% peak HVAC demand reduction and — more relevant to this report's argument — cut the average time from finding a fault to completing the project from 70 days down to 40, while raising the share of identified savings projects that actually got executed from 35% to 70%. That is not a better algorithm; it is a better workflow wrapped around the same class of finding. Independent data-layer players like Noda, Infogrid, and Aquicore are positioning themselves as the neutral execution layer for owners who do not want to hand their entire stack to one controls incumbent — the flip side of the portability argument this series covered when Brick Schema and RealEstateCore began harmonizing their metadata models in June.
Why this matters more in 2026 than it did in 2023
The penalty math has changed the cost of a slow queue. More than 40 U.S. cities now have building-emissions performance standards in effect or scheduled, and per-ton penalties are not static: NYC's Local Law 97 fines run $268 per metric ton of CO2, and Boston's BERDO 2.0 charges $234 per ton plus $1,000 per day of non-compliance, with average penalties rising 82% between compliance periods. A fault sitting in a work-order backlog for 70 days is no longer just lost energy spend — in a growing number of jurisdictions it is compounding regulatory exposure. That reframes the FM's execution rate from an operations KPI into a compliance-risk metric a CFO should be tracking.
| Signal | 2023-era baseline | 2026 data point | What to ask your vendor |
|---|---|---|---|
| FM time on admin work | Assumed "significant" | 61% on work orders + invoices (VTS 2025, n=400) | "Show me the dispatch-to-close-out time, not just the fault count." |
| Savings execution rate | Rarely measured | 35% → 70% with a workflow layer (Noda + LBL field study) | "What % of flagged faults closed in the last 12 months?" |
| Time-to-fix | Not tracked as a KPI | 70 days → 40 days (same study) | "What is our median fault-to-close time today?" |
| Vendor structure | Point diagnostic tools, sold standalone | Diagnostics folded into controls incumbents (Trane/BrainBox AI, JCI/Nantum AI) | "Who physically executes the fix — you, or a subcontractor you dispatch?" |
| Compliance exposure | Flat per-ton fines | Penalties up 82% period-over-period (LL97 / BERDO 2.0) | "What is the $-per-day cost of a fault we don't fix this quarter?" |
Here's what I'd do if this were my building
Before renewing or signing a BaaS/BMS-analytics contract this year, I would ask for one number the vendor almost never volunteers: the trailing-12-month fault-to-close execution rate, not the fault-detection count. A platform that finds 500 issues and closes 150 of them is a worse investment than one that finds 200 and closes 170 — and today's contracts are still priced and marketed on the first number. I would also treat the M&A wave as a buying signal, not just industry noise: if your current vendor is a standalone diagnostics tool with no dispatch arm and no technician network, ask directly whether they plan to partner with a controls incumbent for execution, or whether that gap is yours to fill internally. Either answer is fine — an unanswered question is not. And I would connect the execution-rate conversation to your building's specific compliance calendar (LL97, BERDO, or your local equivalent): a 40-day fault backlog costs a fundamentally different amount depending on whether your penalty period closes in 60 days or 18 months.
The diagnostic layer of this industry solved its problem. The next 18 months of BaaS vendor consolidation are about who owns the layer underneath it — and every FM signing a renewal this year is, whether they've framed it this way or not, buying a position in that fight.
Have a question about this topic? Ask our CRE AI Agent →
Related detections
Related detections — each is a live AISB service module that catches the failure mode above in production: KPI-Theater Detection · EVM-Theater Detection · Retrofit Compliance Scan.
Want these landing in your inbox? Subscribe to The Intelligent Building Brief.