Roadshow
AI Solutions Innovation Roadshow
Coming to a city near you in 2024
News releases
University of Pisa relies on Vertiv for data infrastructure capacity expansion to support high-performance computing and AI
Read the news releaseAI imperatives
Through our Vertiv User Experience process, that integrates in-depth customer interviews, extensive technical proof-of-concepts, and technical collaboration with leading technology partners, Vertiv has curated a collection of key AI infrastructure imperatives that current and future designers, developers and operators of AI factories share:
…be
Transformative.
Operators are embarking on comprehensive changes in leveraging AI technology to transform their products, services and interact with their customers. Facilitating this transformation requires change in strategy, operating model, process, and digital infrastructure, posing several critical infrastructure challenges
Critical infrastructure challenges:
- Retrofitting existing data center infrastructure in a transformative way.
- Rack power density >100kW and accommodating >2268kg
- Liquid and hybrid air-liquid cooling in the data center.
- Liquid distribution as critical as power distribution.
- Power availability and intelligent grid interaction.
…be
First.
Bringing innovative and highly differentiated product, services, and customer experiences first to market.
Critical infrastructure challenges:
- Lengthy design cycles when designing power and cooling paths independently and not as a single system.
- Multi-vendor qualification, dependencies, and technical specifications can vary greatly.
- New technologies and market entrants that lack scalability and global supply chain.
- Significantly more field installation work that is costly, time-consuming, non-repeatable.
- Accelerating complexity and time in commissioning time.
…be
Efficient.
Efficient deployment of capital when retrofitting existing data centers and building new for AI use is critical to improve cost competitiveness and financial sustainability.
Critical infrastructure challenges:
- Robust technical design experience to maximize leverage of existing infrastructure investments.
- Constraints in delivering operational efficiency with densities accelerating >33x.
- Avoiding over-provisioning and stranded capacity when considering fault tolerance.
- Blending new and existing technologies without common language and controls.
- Deploying and maintaining AI factories anywhere in the world.
…be
Confident.
Taking risks and pushing performance to the fullest is critical in delivering defensible differentiation, but doing so in a calculated and intentional manner, avoiding unnecessary and unknown risk-taking.
Critical infrastructure challenges:
- Deep understanding of the technically possible when combining new and existing technologies.
- Getting the most from existing infrastructure investments when retrofitting for AI.
- Planning today's AI factory with future transformations in mind as densities continues to accelerate.
- Robust service and maintenance network with experience, footprint, and trusted performance.
- Designing to use cases with different risk strategies in mind, i.e. Training vs. Inference
…be
Future-Ready.
The past was about tech refresh and data center updates, today is about infrastructure transformations and significant upgrades to support densities in excess of 100kW per rack.
Critical infrastructure challenges:
- Future-proofing the infrastructure investment, mitigating costly future upgrades.
- Designing for power and cooling scalability that can leap, not just grow.
- Consider sustainability that matter today but exponentially grow in scope in the future.
- Partnering with technically and financially viable players that can support into the future.
- Partnering with players with robust ER&D investments driving innovation and closely partnered with industry technology leaders.
Demystifying AI in data centers
Transformation is underway
While still early in its evolution, the impact of AI is already being felt across every industry. From finance and healthcare to manufacturing and government, AI uses cases are proliferating at an extraordinary pace.
Explore the impact of AI across key sectorsAI's economic and technological trajectory
IDC forecasts global AI IT spending to jump from €161.668 billion in 2023 to €468.612 billion in 2027, growing at a rate of 30.4% annually. Spending on AI platforms will continue to rise as they become essential for businesses' digital operations, making cloud technology a critical but supporting component of business transformation.
Analyst report: Global data on AI’s impact on data centers
With the explosion in the use of generative AI transforming operations across every industry, GlobalData analyzed the market and forecast global enterprise AI spend to skyrocket. Discover what that growth means for data centers in their report.
Download the reportThe future of AI infrastructure
Rack density growth is taking off
Accelerated IT achitectures designed to run AI workloads are rapidly driving up the density of data center racks.
*The rack installations split by density data use rack shipments as reference and the average kW/rack is pre-determined for each density range category (i.e. <10kW, 10-20kW, etc.) to help determine the average rack density for each year.
Challenges to data centers
One thing should now be clear: the adoption of AI will not be business as usual for data designers and operators. Dense, power-hungry AI racks can stress virtually every data center system, especially legacy power and cooling infrastructure.
Vertiv™ AI data center infrastructure design principles
Design both power and cooling together to optimize AI infrastructure.
- Power, cooling, and AI hardware compete for limited space and energy.
- A holistic power and cooling design approach is required to maximize the share of space and energy dedicated to AI processing.
Power is at a premium. Eliminate stranded power by aligning AI clusters to data center ‘capacity blocks.
- Power into a data center is segmented into capacity blocks, commonly 1-3 MW and determined by industry-standard sizing of breakers or generators.
- AI is deployed in clusters, soon to be common at 100+ kW / rack, and upward from there.
- Aligning clusters to capacity blocks can ensure that every available kW can be utilized.
Consider total cost of ownership, redundancy, and blast radius in AI power and cooling designs, and the tradeoffs among them.
- The value of AI hardware, at €920,000-3M+ per rack, and the processing it supports, is driving increased consideration of redundancy in power & cooling designs, especially for inference applications.
- Designs that limit blast radius, or the impact from loss of a single capacity segment (server, rack, row), tend to use higher counts of smaller components, potentially at the expense of total cost of ownership.
- Designs that favor total cost of ownership tend to use lower counts of larger components, often with redundancy to reduce the possibility of a loss of a capacity segment.
Handle AI workload surges through system-level controls and power and cooling buffers.
- AI ‘training’ tends to drive large numbers of processors to act in unison, creating massive power consumption surges which can repeat and degrade both the performance and lifespan of power and cooling infrastructure.
- Designs for mitigation include system level controls with rapid response, plus immediately accessible buffers in power and cooling capacity.
Design for a mix of liquid and air cooling, which have a combined and interdependent impact on the ability to remove heat.
- Power into the data center equals the heat rejected.
- Air and liquid cooling temperatures and flows must stay within the operating envelope of both the AI servers and the data center heat rejection equipment.
Design for the future. AI power density will rapidly grow toward 500kW per rack.
- The typical data center life span is almost two decades. The AI chip design cycle is less than two years.
How to keep AI cool
Every step of the cooling process, from chip to reuse, must be carefully planned and modeled to optimize temperature, airflow, and energy efficiency.
Manage the ~5x increase in heat output from AI.
Heat removal from chip to reuse
Effective thermal management in high-density environments requires a holistic approach that encompasses the entire thermal chain – from chip to reuse. Vertiv™ solutions and expertise extend across the thermal chain, providing multiple technologies and design options at each link to tailor a complete solution to site and operator specifications, goals, and constraints.
Liquid cooling solutions
Chilled water cooling solutions
Chilled water systems provide high-efficiency air cooling that is ideal for large data centers. These systems can be configured to handle the full data center load or provide air cooling support for liquid cooling. Heat is rejected through a chiller outside the data center, which also supplies the water used by cooling units.
Explore Vertiv™ chilled water solutions
Direct expansion (DX) cooling
DX cooling systems provide a flexible and reliable approach to cooling the entire data center or supporting liquid cooling. In a DX system, refrigerant absorbs heat from the air and then moves it to an outside condenser, which releases the heat to the atmosphere.
Explore Vertiv™ DX solutionsPowering AI
Understanding the challenge of AI integration
Most data centers today weren’t designed to support the high power requirements of AI racks or effectively remove the heat they generate. The challenges posed by AI integrations can be significant, but they are solvable.
AI workloads are changing how IT is powered, increased CPU power workloads drive higher voltages in the data room all while increasing complexity and constraint on power grids.
Data center power train for AI
An integrated system that promotes maximum uptime, offering reliable back up power to safeguard against outages and maintain continuous operations.
Explore Vertiv™ data center power systems for AI
Vertiv™ dynamic online mode UPS
Designed for high-density computing environments, offering up to 99% operational efficiency without sacrificing power protection.
Transform your power management strategy
Intelligent high-density power distribution
Not all rPDUs are created equal. Vertiv™ Geist™ Switched Rack PDUs provide the visibility, efficiency, and reliability required to support complex workloads.
Battery Energy Storage Systems (BESS)
BESS with lithium-ion batteries enable a steady power supply and reliable backup for demanding tech environments.
Enhance your power infrastructure with BESS solutions
Open busbars
Reduce cable clutter and complexities associated with raised floors, enhance safety, cut costs, and boost efficiency for critical computing workloads in data centers.
Explore how open busbars work and their benefitsServices keeping AI running
End-to-end solutions and services
Vertiv offers the industry's most robust line of service offerings for data center liquid cooling systems. Our trained technicians have expansive knowledge on how the systems operate and integrate with AI and other IT equipment to deliver relief to new liquid cooling adopters. Vertiv's holistic approach to liquid cooling services and maintenance helps ensure systems function properly and high-density, high-value racks are cooled efficiently.
Explore Vertiv's full line of design, installation and maintenance services
Transition assessment
Transition to liquid cooling with confidence by scheduling a site assessment from Vertiv. Our thermal experts can provide a roadmap for upgrading your infrastructure for AI-readiness, ensuring a seamless transition and enhancing operations. Get expert recommendations on best practices for future reliability and risk mitigation to prepare your infrastructure for the AI revolution.
Lifecycle services
Vertiv's lifecycle services provide continuous insight into critical equipment operation to improve performance, streamline service processes for faster issue resolution, and add deep equipment and service expertise to any organization.
Explore Vertiv's service capabilitiesVertiv™ 360AI end-to-end power and cooling solutions for AI
The simple way to power and cool AI
Vertiv™ 360AI includes a complete portfolio of power, cooling, and service solutions that solve the complex challenges arising from the AI revolution. Within the Vertiv 360AI solutions portfolio, Vertiv has developed reference designs for retrofit and greenfield data centers to help you navigate integrated solutions for power and cooling AI and other high-performance computing applications.
Click on the designs below based on your needs to see our recommended design for your situation. Be sure to click the “Talk to an Expert” to discuss further with one of our application engineers.
Design for new builds
| Rack density | Rack count | GPU count | Design ID | Cooling technology | |
|---|---|---|---|---|---|
| North America | EMEA | ||||
| 20kW | 18 | 248 | RD002 |
RD002E |
Air |
| 40kW | 10 | 248 | RD003 |
RD003E |
Air |
| 40kW | 10 | 248 | RD004 |
RD004E |
Air |
| 73kW | 88 | 2304 | RD006 |
RD006E | Liquid + Air |
| 73kW | 110 | 2880 | RD007 |
RD007E | Liquid + Air |
| 132kW | 36 | 1152 | RD014 |
RD014E | Liquid + Air |
| 132kW | 54 | 1728 | RD015 |
RD015E | Liquid + Air |
| 132kW | 72 | 2304 | RD016 |
RD016E | Liquid + Air |
| 300kW | - | - | RD300 | RD300E | Liquid |
| 500kW | - | - | RD500 | RD500E | Liquid |
Design optimized for retrofits
| Rack density | Rack count | GPU count | Design ID | Cooling technology | |
|---|---|---|---|---|---|
| North America | EMEA | ||||
| 40kW | 4 | 128 | 4X160R |
4X160RE | Air |
| 70kW | 1 | 64 | 1L70R |
1L70RE |
Liquid + Air |
| 100kW | 1 | 88 | 1L100R |
1L100R |
Liquid + Air |
| 100kW | 4 | 368 | 4L400R |
4L400RE | Liquid + Air |
| 100kW | 4 | 368 | 4XL400 |
4XL400 |
Liquid + Air |
| 100kW | 5 | 460 | 5L500 |
5L500 |
Liquid + Air |
| 100kW | 12 | 1104 | 12XL1200 |
12XL1200 |
Liquid + Air |
| 100kW | 14 | 1288 | 14L1400 |
14L1400 |
Liquid + Air |
Benefits of Vertiv™ 360AI solutions
Reduce Deployment
Time by up to 50%
Pre-engineered solutions simplify the deployment process end-to-end and eliminate design cycles, reduce deployment time up to 50%.
Broad Range and Flexible
Deployment Methods
Solutions available from Edge Inferencing to AI Data centers with options ranging from a high-density rack solution, to large prefabricated modular data centers
Interoperability
and Seamless Operation
Only Vertiv can meet both the power and cooling demand of AI, and each solution is built from the most complete portfolio of power and cooling in the industry.
Unmatched Expertise
& Support
Support throughout the process from assessment and design, to deployment and lifecycle management
Accelerate your AI deployment
Speed and simplicity throughout the deployment is crucial for enterprises looking to deploy AI and high-performance computing.
Pre-engineered solutions are proven to dramatically reduce design cycles, reduce risk, and provide more predictable costs.
Build my AI solutions
Vertiv's AI solutions
Leading AI chip makers partner with Vertiv
Adding Vertiv Expertise to the NVIDIA Partner Network
Vertiv and NVIDIA collaborated to develop a unique AI cooling solution for the COOLERCHIPS program and formalized our partnership when Vertiv joined the NVIDIA Partner Network. NVIDIA’s cutting-edge platforms need advanced infrastructure solutions. AI deployments can count on Vertiv’s leadership in power and cooling to keep their AI running no matter the cluster size.
Discover the future of AI with Vertiv and NVIDIA
Supporting Intel’s Latest Generation AI Accelerator
Vertiv is collaborating with Intel to provide liquid- and air-cooled solutions for the revolutionary Gaudi3 AI accelerator. Our liquid cooled solution has been tested at 160kW accelerator power using facility water from 17°C to 45°C (62.6°F to 113°F). The air-cooled solution has been tested up to 40kW of heat load that can be deployed in warm ambient air data centers up to 35°C (95°F).
Dive into the Vertiv-Intel collaborationTalk to an expert
Fill out the details to have our AI solutions consultants get back to you.
Your submission has been received!
Thank you for reaching out to us.
Learn more
Unlock deeper AI insights and stay ahead of the curve
How to implement liquid cooling at existing data centers
Read our technical guide for practical strategies and a step-by-step roadmap for adopting liquid cooling for 1MW IT loads.
Read more
Join us in one of our upcoming events
July 11, 2024
AI solutions innovation roadshow
Register today to learn how Vertiv’s comprehensive AI-ready portfolio powers and cools even the highest-density workloads in the most challenging deployments.
Register now
June 4-6, 2024
Datacloud Global Congress & Awards 2024
We are looking forward to meeting you in Cannes from 4th to 6th of June, at the prestigious industry event Datacloud Global Congress.
Register now
October 9-10, 2024
World AI Summit, 9-10 October 2024, Amsterdam
The AI industry is standing on the brink of a seismic shift that threatens to shake the very foundation of our technological existence.
Register now
