Wirewalk Instituteof AI
inf-340 Expert Waitlist

Data Center Power and Liquid Cooling for AI

For data center, facilities and infrastructure engineers who must host high-density GPU racks and leave able to design, commission and operate the power and liquid-cooling systems they need.

The curriculum is being prepared and reviewed. Enrollment is not open yet.

About this course

An 8-GPU air-cooled HGX-class server draws roughly 10 to 15 kW; rack-scale systems such as GB200 and GB300 NVL72 bring a single rack well past 100 kW and require liquid cooling, and vendor roadmaps point toward racks near a megawatt. Facilities designed for 8 to 15 kW per rack cannot simply absorb that. This course teaches the electrical and thermal engineering behind AI racks so infrastructure and facilities teams can plan, build and run them safely.

The power modules trace the chain from utility and medium voltage through UPS, switchgear, busway and PDUs to rack power shelves, then examine GPU power behavior: fast load swings during training, power capping and smoothing, and the redundancy schemes that fit AI loads. You study the industry move from 54 V in-rack distribution toward 800 VDC architectures announced by NVIDIA and its partners. The cooling modules cover heat transfer basics, direct-to-chip liquid cooling with coolant distribution units, technology cooling system and facility water loops, ASHRAE's W-class and S-class temperature definitions, fluids, filtration and leak detection, then immersion and rear-door heat exchangers, heat reuse and water use. The course closes with retrofit strategies, commissioning and integrated system testing, and day-2 operations including monitoring, maintenance and incident response. Facts are stated as of October 2026; this course does not replace licensed engineering design.

What you will be able to do

  • Trace the electrical chain from utility to GPU and calculate rack, row and hall power for a stated AI deployment
  • Explain GPU power dynamics during training and choose power capping, smoothing and redundancy strategies accordingly
  • Explain the rationale and trade-offs of 800 VDC distribution versus 54 V in-rack and AC architectures
  • Design a direct-to-chip liquid cooling loop with CDUs and specify facility and technology cooling system temperatures using ASHRAE classes
  • Compare direct-to-chip, immersion, rear-door heat exchanger and air cooling for a given density and building
  • Plan a retrofit of an air-cooled hall for liquid-cooled AI racks, including structural, piping and phasing constraints
  • Write a commissioning and integrated systems test plan and a day-2 operations runbook for liquid-cooled racks

Syllabus

6 modules · 19 lessons · about 18 hours

Module 1: Why AI Breaks the Data Center

The density step change and the arithmetic every facilities conversation starts from.
3 lessons
  1. From 10 kW to Rack-Scale: How GPU Density Changed Facility Design
  2. Power and Heat Arithmetic: kW, BTU, Flow Rate and Delta-T
  3. PUE, WUE and What Efficiency Metrics Hide

Module 2: The Power Chain

Delivering reliable power from the grid connection to the GPU.
3 lessons
  1. Utility, Medium Voltage, Switchgear and UPS Topologies
  2. Busway, PDUs and Rack Power Shelves in ORv3 and NVL72-Class Racks
  3. 800 VDC Distribution: Architecture, Safety and Migration Path

Module 3: GPU Power Behavior

How AI loads differ electrically from traditional IT and how to control them.
3 lessons
  1. Synchronous Load Swings in Training and Their Effect on Upstream Equipment
  2. Power Capping, Power Smoothing and Energy Storage
  3. Redundancy Choices: N, N+1, 2N and Distributed Redundancy for AI

Module 4: Direct-to-Chip Liquid Cooling

Cold plates, CDUs and the water loops that carry heat out of the building.
4 lessons
  1. Cold Plates, Manifolds and Quick Disconnects
  2. CDUs, Technology Cooling System Loops and Facility Water
  3. ASHRAE W-Class and S-Class Temperatures and Choosing a Supply Temperature
  4. Fluids, Filtration, Water Quality and Leak Detection

Module 5: Alternatives and Retrofits

Other cooling approaches and how to bring liquid cooling into an existing hall.
3 lessons
  1. Immersion Cooling: Single-Phase and Two-Phase
  2. Rear-Door Heat Exchangers, Hybrid Air-Liquid Halls and Heat Reuse
  3. Retrofitting an Air-Cooled Hall: Structure, Piping, Phasing and Risk

Module 6: Commissioning and Operations

Proving the facility works under real load and keeping it working.
3 lessons
  1. Commissioning Levels and Integrated Systems Testing with Load Banks
  2. Monitoring Power and Coolant: BMS, DCIM and Telemetry from the Rack
  3. Day-2 Runbooks: Maintenance Windows, Coolant Events and Power Incidents

Exam and certificate

The exam unlocks once you have completed every lesson. It draws 25 questions from a larger bank, spread across the modules, with the answer options shuffled for each attempt. You have 60 minutes and need 80% to pass. After a failed attempt you can try again after 24 hours, up to 3 attempts in total. Feedback is given per module, never as the answer key.

Passing earns a certificate of completion and assessment from the Wirewalk Institute of AI with a unique ID that anyone can check at institute.wirewalk.com/verify. It is not an accredited qualification or a vendor certification. How certification works.

The Wirewalk Institute of AI is independent. This course teaches publicly documented products and is not affiliated with or endorsed by NVIDIA, Open Compute Project, ASHRAE, Vertiv, Schneider Electric, Eaton, or CoolIT Systems. Product and company names are trademarks of their respective owners.

More in AI Infrastructure and HPC

All 14 courses in this track
inf-101 Basic Waitlist

AI Infrastructure Explained

A plain-language tour of the stack that runs AI: what a GPU server is, why clusters need special networks and storage, how jobs are scheduled, what an AI data center needs in power and cooling, and how cloud and on-premises options differ. No prior infrastructure background required.

6 h · 15 lessons Join the waitlist
inf-201 Intermediate Available

GPU Infrastructure for AI

The operator's view of data-center GPUs: what the current NVIDIA and AMD systems are, how they are packaged and connected, how the driver, CUDA/ROCm and container stack fit together, how to monitor and triage failing GPUs, how to share and size GPU memory, and what power, cooling and sourcing decisions follow.

12 h · 16 lessons See price at checkout
inf-210 Intermediate Waitlist

Linux Administration for AI Clusters

The Linux skills that differ on an AI cluster: choosing and imaging the OS, packaging GPU and RDMA drivers that survive kernel updates, NUMA and PCIe topology tuning, cgroup v2 isolation, shared identity and file systems, scientific software environments with Lmod, Spack and containers, and patching a fleet without breaking jobs.

14 h · 19 lessons Join the waitlist