Live from VRT Analyst Day - Part 3 (Chief Product & Technology Officer Comments)

Research evidence excerpt

Live from VRT Analyst Day - Part 3 (Chief Product & Technology Officer Comments)

system cost for customers and higher total Vertiv content, and pushes

800 VDC downstream via the PowerDirect 5000 sidecar (up to 900 kW,

enabling ~300 kW–1 MW+ per rack; commercial launch early 2027).

Stage 3, for purpose-built large-scale AI factories, moves to centralized

DC distribution in the data hall via two parallel paths: a near-term MV

DC UPS using mature transformer/rectifier tech, and a longer-term

solid-state transformer path for greater density and efficiency. Critically,

these three architectures will coexist rather than replace each other and

Vertiv’s role and value increase across every stage of AI power

architecture evolution.

Thermal Management: Vertiv's system-level approach manages the

entire liquid loop (CDU, secondary fluid network, chip/server cooling,

fluid management services, and Unify software for continuous

optimization and leak detection) with new CDU platforms scaling

beyond 2 MW. The new product is the CoolLoop Trim Cooler, which

combines dry-cooler efficiency with chiller backup and claims up to 79%

lower annual energy consumption, a 24% smaller footprint, up to 134%

more annual free-cooling hours, zero water use, and scalability above

3 MW across current and future GPU generations. Architecturally,

Vertiv positions both dual-loop designs (separate, independently

optimized air and liquid loops for maximum efficiency) and optimized

single-loop designs using CoolPhase Water-Cooled DX (simpler

footprint and lower TCO), letting customers pick by efficiency vs.

deployment-simplicity priorities’ portfolio is deliberately broad so the

optimal architecture can be selected by ambient temperature, GPU

type, peak power, footprint, and risk profile.