HQVT is bringing multispectral AI deeper into data-center infrastructure, using ultraviolet, infrared and visible-light sensing to identify early signs of thermal and electrical anomalies. The company’s latest terminals are designed to complement existing DCIM platforms as higher-density AI computing increases pressure on power, cooling and equipment reliability.
As AI workloads push data centers toward higher rack densities and greater power consumption, infrastructure operators are facing a monitoring problem that conventional alarms and periodic inspections may not fully solve. Small changes in temperature, heat distribution or electrical behavior can develop before equipment crosses a predefined threshold.
Shenzhen-based HQVT Technology is addressing that gap with multispectral AI terminals designed to continuously monitor physical conditions inside and around data-center equipment.
The company showcased its HQ 6000 Multispectral AI Terminal, In-Cabinet Multispectral AI Terminal and Privacy-Preserving Multispectral AI Terminal at Data Centre World Asia 2026 in Singapore. The event took place September 29–30 at Marina Bay Sands and focused on AI-ready infrastructure, power, energy efficiency, cooling, safety and data-center operations.
The technology arrives as power density becomes a central constraint on AI infrastructure. Gartner estimates worldwide data-center electricity consumption will reach 565 TWh in 2026, up 26% from 2025. AI-optimized servers are expected to account for 31% of data-center power consumption this year, with their power consumption forecast to exceed that of conventional servers in 2027.
That environment makes physical monitoring more important. A cabinet operating under a heavier AI workload can experience localized heat buildup or electrical abnormalities that may not immediately trigger conventional facility alarms.
HQVT’s approach combines sensing across multiple spectral bands with on-device AI analysis. Its systems are designed to look for changes in equipment conditions rather than relying solely on conventional visible-light video or individual threshold alerts.
The In-Cabinet Multispectral AI Terminal takes that monitoring closer to the source of potential problems. Installed inside electrical or server cabinets, it is designed to identify abnormal heat and electrical-discharge signals that could otherwise remain difficult to observe.
The HQ 6000 provides a broader multispectral sensing configuration for data-center environments, allowing operators to deploy monitoring according to facility layout and equipment density.
The company is also targeting environments where conventional cameras create privacy concerns. Its Privacy-Preserving Multispectral AI Terminal uses ultraviolet and infrared sensing without visible-light imaging. That means it can continuously monitor for anomalies without creating conventional video imagery, potentially making the approach more suitable for privacy-sensitive facilities.
The broader architecture extends beyond individual cabinets. HQVT describes a Three-tier Collaborative Multispectral AI Perception System comprising in-cabinet sensing, spatial sensing across aisles and critical areas, and mobile inspection for locations that fixed sensors cannot easily cover.
The resulting data can be combined with equipment status, historical trends and facility information through HQVT’s Zhiyuan Origin Large Model. The company says the platform supports complex-event correlation, confidence scoring and risk grading.
This architecture reflects an important change in how AI can be used in data-center operations. Instead of treating AI as a replacement for data-center infrastructure management (DCIM), HQVT is positioning multispectral AI as an additional physical-observation layer that feeds richer information into existing management systems.
That distinction could become increasingly relevant as AI data centers become more complex. The International Energy Agency estimates data-center electricity consumption was about 415 TWh in 2024 and projects it could reach around 945 TWh by 2030 in its base case. Accelerated servers, driven largely by AI, are expected to account for almost half of the net increase in global data-center electricity consumption.
More compute therefore means not only more power and cooling infrastructure, but also more opportunities for localized physical failures to affect increasingly expensive workloads.
The competitive landscape includes conventional DCIM, thermal monitoring, electrical monitoring, computer vision and predictive-maintenance technologies. Multispectral sensing represents a more specialized approach, with its potential advantage coming from combining information that individual sensing modalities may miss.
For operators, the practical question will be whether this additional layer can reduce false alarms, identify problems early enough to prevent failures and integrate cleanly with existing facility-management systems.
HQVT’s Southeast Asian expansion comes as regional data-center operators prepare for continued AI-driven capacity growth. The company’s strategy is less about replacing existing infrastructure software and more about giving those systems another source of physical intelligence.
As AI infrastructure moves toward higher power densities, continuous sensing could become an increasingly important component of data-center resilience. The value of multispectral AI will ultimately depend on whether it can turn subtle physical signals into actionable warnings before a thermal or electrical anomaly becomes an operational incident.
Market Landscape
AI is increasing both the density and energy requirements of data-center infrastructure. Gartner forecasts global data-center electricity consumption of 565 TWh in 2026, with AI-optimized servers responsible for 31% of consumption.
The resulting infrastructure challenge extends beyond GPUs and cooling. Operators need better visibility into physical conditions at rack, cabinet, aisle and facility levels. DCIM platforms already aggregate operational information, but emerging AI-based sensing systems can add continuous physical observations to those datasets.
HQVT’s approach sits at the intersection of AI infrastructure, predictive maintenance, computer vision alternatives, data-center safety and intelligent facility management. Its privacy-preserving sensing is particularly relevant for environments where continuous monitoring is required without conventional video capture.
Top Insights
- HQVT’s multispectral AI terminals add continuous physical sensing to data-center environments, targeting thermal and electrical anomalies before they become equipment failures.
- The in-cabinet configuration moves monitoring closer to high-density computing equipment where localized heat and electrical abnormalities can develop.
- Ultraviolet and infrared sensing enables anomaly detection without conventional visible-light video, addressing privacy concerns in sensitive environments.
- HQVT’s three-tier architecture combines cabinet, spatial and mobile sensing to create broader visibility across complex data-center facilities.
- The technology is positioned as a complementary layer for DCIM rather than a replacement, feeding additional physical intelligence into existing operational systems.
Power Tomorrow’s Intelligence — Build It with TechEdgeAI
