The scalable solution supports deployment in existing enterprise facilities, new builds, commercial spaces, colocation and hyperscale environments without redesign
Globally recognized as a critical digital infrastructure and continuity solutions innovator, Vertiv Holdings Co. introduced the Vertiv™ SmartIT solution. The platform is a pre-validated, plug-and-play architecture that addresses density and integration pain points for enterprises, high performance computing centers, and colocation providers expanding high-density AI compute in both greenfield and brownfield deployments across existing real estate.
Designed in partnership with NVIDIA and first launched to support NVIDIA DGX™ GB300, the world’s highest performance enterprise AI system, the Vertiv™ SmartIT fully integrated stack incorporates all elements of a solution – enterprise server racks, direct to chip liquid cooling, secondary fluid distribution, DC power conversion, and integrated telemetry – into a single pre-commissioned footprint that eliminates multi-system interoperability issues prior to purchase and requires only the distribution system hookup to commission on site. The platform is now available immediately worldwide.
“As the complexity of data center workloads continues to grow, AI will drive fundamental changes in how data center infrastructure is provisioned and operated with more synergy needed among power, cooling and IT,” said Scott Armul, Chief Product and Technology Officer, at Vertiv. “Vertiv™ SmartIT embodies our philosophy on this challenge: delivering this synergy through a factory-validated architecture. By integrating our knowledge across these disciplines, we’re enabling customers to streamline deployment, improve consistency in their AI deployments and create a platform for future growth.”
Removing core deployment bottlenecks for accelerated computing
The Vertiv™ SmartIT platform is the immediate cure for many of these operational pain points and supply chain frustrations that have historically been holding back the proliferation of dense compute:
- Faster Time-to-Compute: In-plant testing of the modules reduces significantly the time it takes to reach commissioning onsite, with only the basic fluid and power connections.
- Power, thermal and management. Companies are focusing on their sources of competitive advantage and outsourcing others; companies collaborate around a sourcing, design and engineering partner in development, manufacturing and post-sales support of a single partner and integrated architecture.
- Brownfield and Commercial Readiness Legacy enterprise data center footprints and high-tier commercial office buildings enable provision of compute intensive AI clusters, with no significant renovation or costly reconfiguration of flooring.
- Repeatable, Linear Scalability: Using the same engineering specifications, operators may scale from first isolated racks to multi-megawatt rows.
- With embedded Operational Safety fluid leak detection integrated monitoring and active mitigation routines from the individual chassis back to the facility manager, rather than a patchwork third-party monitoring stack.
Also Read: Lattice Semiconductor and Arm Partner to Drive Secure, Adaptable Control Platforms for Next-Generation AI Data Centers
Broad Architecture Support and Integrated Ecosystem
Vertiv™ SmartIT enables high-density accelerated architectures built on NVIDIA MGX™ and Open Compute Project (OCP) Open Rack Wide (ORV3) standards, with turnkey support for NVIDIA GB300 NVL72, NVIDIA Vera Rubin NVL72, NVIDIA DGX™ Rubin NVL8, and NVIDIA HGX™ Rubin NVL8 platforms.
This solution natively supports existing data center management standards such as Redfish, Modbus, and SNMP and is supported by Vertiv™ Global Services for local project commissioning, emergency dispatch, and asset life-cycle management.
In general enterprise deployments, Vertiv™ SmartIT integrates with other components of the portfolio. Several partners have worked with Vertiv for many years to deploy critical infrastructure and data center solutions across a variety of physical environments and challenges. Vertiv™ Holdings, a global manufacturer of critical digital infrastructure and continuity solutions, announced today the launch of the Vertiv™ SmartIT solution. This pre-tested, one-click, plug-and-play architecture addresses density and integration issues for enterprise, high-performance computing data centers and colocation deployments that are expanding high-density AI compute within their current real estate or new greenfield locations.
Created with NVIDIA and first launched to power the industry’s leading enterprise AI system, the award-winning NVIDIA DGX™ GB300-Vertiv™ SmartIT is a fully integrated, holistic infrastructure stack. That stack-the enterprise server rack, direct-to-chip liquid cooling, secondary fluid distribution, direct current power conversion, and telemetry-is factory integrated to come online as a single, pre-commissioned footprint. No multi-system interoperability is required before delivery, which reduces on-site commissioning to the facility hookup, delivering production AI to the market faster. The platform is immediately available in markets worldwide.
“AI is forcing a new level of integration between data center systems because it increases the need for IT, cooling and power systems to operate in a more coordinated fashion to accommodate more complex workloads,” said Scott Armul, chief product and technology officer, Vertiv. “Vertiv™ SmartIT is our response to that requirement by integrating these disciplines into a single, factory-validated platform. This integration allows us to leverage Vertiv’s deep expertise in critical infrastructure to help our customers simplify deployment, reinforce the consistency of their AI operation, and create an infrastructure foundation that can adapt to the future of their computing needs.”
Eliminating Core Deployment Bottlenecks for Accelerated Computing
The Vertiv™ SmartIT platform directly addresses the operational and supply chain issues that have created obstacles to deploying dense compute in the past:
- Fast Time-to-Compute: Manufacturing-level validation greatly reduces on-site commissioning to just fundamental fluid and power connections.
- Management Companies manage sourcing, design and engineering, and lifecycle support of a single partner and integrated architecture, with power, thermal, and management.
- Commercial and brownfield Readiness: Existing enterprise datacenter footprints and high-tier commercial buildings enable compute-intensive AI clusters to be built without extensive renovation or costly reconfiguration of flooring.
- Repeatable, Linear Scalability: Operators can leverage the same engineering inputs as they scale from hundreds of kilowatts in initial isolated racks through multiple megawatt rows.
- Unified systems that include integrated fluid leak detection, active mitigation routines, and end-to-end monitoring provide a single, unified view from the individual chassis back to the Central Facility Manager, instead of a patchwork third-party monitoring stack.
Extensive Architecture Compatibility and End to End Ecosystem
Vertiv™ SmartIT: Accelerated high-density architecture Accelerated high-density platforms are supported by both the NVIDIA MGX™ architecture and the Open Compute Project (OCP) Open Rack Wide (ORV3) standard – with a tested ready design that is compatible out of the box with NVIDIA GB300 NVL72, NVIDIA Vera Rubin NVL72, NVIDIA DGX™ Rubin NVL8 and NVIDIA HGX™ Rubin NVL8 platforms.
The system offers native integration with traditional data center management protocols-and is supported by the Vertiv™ Global Services team for local project commissioning, emergency response and asset lifecycle services.
When used in larger enterprise deployments, the Vertiv™ SmartIT solution works in concert with other elements of the portfolio. Customers can utilize Vertiv™ Smart Run for row-level orchestration, Vertiv™ intelligent rack power distribution units (rPDUs) for granular power measurement, and Vertiv™ Unify for centralized operational data collection, thermal closed-loop control, and correlation of faults.
Earlier this year, the company demonstrated the platform’s performance at NVIDIA GTC Berlin, demonstrating an integrated Vertiv™ SmartIT demonstration unit equipped with NVIDIA HGX™ Rubin NVL8 hardware that can provide flexible avenues for enterprise AI expansion.
Help improve row-level coordination by setting up Vertiv™ smart rack power distribution units (rPDUs) with fine-tuned power measurement and Vertiv™ Unify to analyze operations centrally, monitor thermally with a closed-loop system, and coordinate with the fault detector.
Prior to this release, the company demonstrated the platform’s functionality at a previous NVIDIA GTC in Berlin by delivering an end-to-end Vertiv™ SmartIT demo unit built using NVIDIA HGX™ Rubin NVL8 hardware platform that demonstrated scalable growth options for enterprise AI.
