About
Verdigris delivers continuous electrical intelligence for AI data centers, commercial buildings, and industrial facilities. Unlike standard monitoring tools that poll at 1 Hz, Verdigris captures waveform data at 8 kHz, generating 1,600 data points per event to identify specific failure signatures before they cause outages. In a real-world deployment at T-Mobile, the platform detected active degradation in 4% of 800+ rectifiers up to 21 days before failure — all of which had triggered zero standard alarms. Beyond predictive failure detection, Verdigris addresses the capacity crunch facing AI infrastructure operators. Circuit-level, real-time electrical data exposes stranded capacity that aggregate metering misses, with operators typically recovering 15–25% of previously inaccessible capacity. The platform also provides before-and-after measurement and verification (M&V) for every infrastructure change, from GPU rack additions to UPS replacements and cooling optimization. Deployment is fast and non-disruptive: clamp-on current transformers (CTs) install in minutes per panel with zero downtime, and secure cellular backhaul eliminates firewall changes. Verdigris operates on a cloud-first architecture with continuous algorithmic validation to catch sensor drift and data gaps. An API-first design lets teams feed detection events and waveform records into Grafana, ServiceNow, or any data lake via a documented REST API, making it a complementary layer to existing monitoring stacks. It is trusted across 2+ GW of monitored peak demand and 20M+ square feet of global deployments.
Key Features
- 8 kHz Waveform Analysis: Captures electrical data at 8,000 samples per second — 8,000× the resolution of standard 1 Hz polling — to identify subtle degradation signatures that conventional monitoring cannot detect.
- Predictive Failure Detection: Continuously analyzes waveform signatures to flag active equipment degradation up to 21 days before failure, validated in real deployments against fleets of 800+ rectifiers.
- Stranded Capacity Recovery: Circuit-level, real-time electrical visibility reveals unused power capacity hidden from aggregate metering, enabling operators to recover 15–25% of stranded capacity per facility.
- Zero-Downtime Hardware Deployment: Clamp-on CT sensors install in minutes per panel without shutting down systems; secure cellular backhaul eliminates the need for firewall changes or IT network integration.
- API-First Integration: Every detection event and waveform record is accessible via a documented REST API, enabling direct feeds into Grafana, ServiceNow, data lakes, or any existing monitoring stack.
Use Cases
- Detecting rectifier and UPS degradation in AI data centers before it causes unplanned downtime, using continuous 8 kHz waveform analysis.
- Recovering stranded power capacity across multi-site data center fleets by identifying underutilized circuits through circuit-level electrical visibility.
- Validating the power impact of infrastructure changes — such as adding GPU racks or replacing cooling systems — with before-and-after continuous M&V.
- Monitoring large portfolios of commercial or industrial facilities for electrical anomalies using rapid, zero-downtime sensor deployment and cellular backhaul.
- Integrating high-resolution electrical event data into existing NOC dashboards, ITSM platforms, or data lakes via a documented REST API.
Pros
- Detects What Standard Monitoring Misses: The 8 kHz sampling rate captures transient harmonic spikes and degradation patterns that are structurally invisible to 1 Hz polling systems, providing a genuine safety layer for AI workloads.
- Fast, Non-Disruptive Deployment: Clamp-on sensors and cellular backhaul mean hundreds of panels can go live within weeks without downtime, firewall changes, or complex IT coordination.
- Tangible ROI via Capacity Recovery: Recovering 15–25% stranded capacity across large fleets translates directly into deferred infrastructure spend and increased revenue density per facility.
- Integrates With Existing Stacks: The REST API and cloud-first architecture make Verdigris an additive layer rather than a rip-and-replace, working alongside existing DCIM, SCADA, or BMS tools.
Cons
- Enterprise-Only Pricing: Verdigris targets large-scale operators of data centers and industrial facilities; pricing and deployment are not self-serve, making it inaccessible for smaller or experimental use cases.
- Hardware Dependency: Full value requires physical installation of clamp-on CT sensors, which — while fast — means the platform is not purely software and requires on-site coordination.
- Narrowly Focused Use Case: The platform is purpose-built for electrical infrastructure monitoring and is not applicable to general IT, software, or non-power-intensive operational environments.
Frequently Asked Questions
How is 8 kHz monitoring different from standard 1 Hz monitoring?
Standard monitoring polls once per second. A 200 ms harmonic spike from a degrading rectifier is averaged away and appears as a normal reading. At 8 kHz, the same event generates 1,600 data points, enabling Verdigris to identify the specific failure signature and flag it continuously.
How quickly can Verdigris be deployed?
Clamp-on current transformers install in minutes per panel with zero downtime. Secure cellular backhaul eliminates firewall changes. In one deployment, 1,400+ panels at T-Mobile facilities were live within weeks of the deployment start date.
What does 'stranded capacity recovery' mean in practice?
AI workloads demand more power per rack, but aggregate metering often can't identify where unused circuit capacity exists. Verdigris provides per-circuit, real-time electrical data so operators can confidently deploy more load — typically recovering 15–25% of previously inaccessible capacity.
Can Verdigris integrate with my existing monitoring tools?
Yes. Verdigris is API-first: every detection event and waveform record is available via a documented REST API. It feeds directly into tools like Grafana, ServiceNow, or any data lake, and is designed to complement rather than replace existing DCIM or BMS systems.
What industries and facility types does Verdigris support?
Verdigris is purpose-built for AI data centers, commercial buildings, and industrial facilities — any environment with critical electrical infrastructure where unplanned downtime or power inefficiency carries significant operational or financial cost.
