On August 5, 2026, before the U.S. market open, data center infrastructure leader Vertiv released its Q2 fiscal 2026 earnings. The report showed that, thanks to explosive growth in global AI computing infrastructure, quarterly orders surged 40% year over year, backlog hit a record high, and management raised full-year revenue and profit guidance. Following the announcement, Vertiv shares jumped more than 12% in pre-market trading, lifting the entire data center power and cooling equipment sector.
This is not the first time Vertiv has benefited from AI. As AI chip power from Nvidia, AMD and others continues to climb, rack power density has jumped from 10 kW in the traditional air-cooled era to 50-100 kW, and the next-generation NVL72 rack system will exceed 120 kW. The cooling challenge of high power density has made liquid cooling "mandatory" rather than "optional," while power supply efficiency and redundancy have become the biggest headaches for data center builders.
AI Computing Demand Explodes: Power and Cooling Move from Supporting Role to Center Stage
In the past, attention on AI computing focused almost entirely on GPU chips and HBM memory. In fact, every $1 invested in AI server chips requires about $1.2 or more in infrastructure. Power systems, cooling systems, rack busways and backup batteries are becoming the fastest-growing capex segments in the AI industry chain.
In Vertiv's Q2 report, data center solutions revenue grew 38% year over year, with liquid cooling CDU (coolant distribution unit) shipments doubling quarter over quarter, and digital service subscription revenue from management software and monitoring platforms rising 45%. The CEO said on the conference call: "As AI clusters expand from a few thousand GPUs to tens or even hundreds of thousands of GPUs, power and thermal management have become core variables determining project delivery cycles and costs. Not only North American cloud giants but also European sovereign clouds and Middle Eastern sovereign wealth funds are accelerating adoption of high-density liquid cooling."
Vertiv Earnings Highlights: Soaring Orders and Gross Margin Pressure
- Quarterly orders reached $5.2 billion, up 40% year over year, a record high;
- Revenue was $2.48 billion, up 29% year over year, above market consensus;
- Gross margin was 36.8%, down 1.2 percentage points year over year, mainly due to higher logistics and key raw material costs;
- Backlog stood at $15 billion, equivalent to 2.5 quarters of future revenue.
By business mix, Vertiv's traditional businesses (telecom and commercial buildings) stayed stable, while data center business rose to 72% of revenue, becoming the absolute core. Management guided to 22%-24% organic revenue growth for the full year, up 4 percentage points from the previous forecast. Although gross margin faces near-term pressure, scale effects and high-value-added data center services are expected to improve in the second half.
Liquid Cooling Technology Race: Cold Plate Becomes Mainstream, Immersion Ready to Surge
AI chip power consumption keeps rising. Nvidia's Blackwell Ultra and Rubin series GPUs have TDPs evolving from 700W to 1500W, and traditional air cooling clearly cannot meet cooling requirements. Cold plate liquid cooling currently accounts for about 70% of new AI data centers thanks to its technical maturity and cost-effectiveness, with Vertiv, Schneider Electric and Delta Electronics as main beneficiaries.
Meanwhile, immersion liquid cooling is more efficient but has high retrofit costs, and is currently used mainly in national supercomputing centers and a few hyperscale customers. Startups such as Verdant and GRC continue to push it, and Vertiv has also launched a single-phase immersion solution based on fluorinated fluid as a technology reserve. The industry expects immersion cooling penetration in AI data centers to reach 30% by 2028.
AI Data Center Power Bottleneck: From 'Power Availability Review' to 'Microgrid'
Besides cooling, power supply is a bigger bottleneck. In U.S. data center hubs such as Northern Virginia and Montana, grid interconnection queues have exceeded three years. AI clusters can require hundreds of megawatts, forcing many projects to build natural gas or storage microgrids.
Vertiv specifically noted in its earnings that orders for modular power and intelligent power distribution products are growing rapidly, and its 48V power distribution architecture for high-density racks has been certified by several cloud vendors. The company also launched an integrated lithium-battery UPS system for AI campus-level backup power to meet emergency needs lasting one hour or more.
Win-Win Across the Supply Chain: Who Is the Next Beneficiary?
Vertiv's strong performance is a microcosm of the AI infrastructure boom. Along the supply chain, the following directions deserve attention:
- Power equipment makers: Electrical equipment leaders Eaton (ETN), Rockwell Automation (ROK) and others also benefit from data center power expansion;
- Thermal management and liquid cooling components: Boyd (BOYD), Corning (GLW) and others are involved in the liquid cooling chain through different products;
- Semiconductor equipment and cybersecurity: AI data center construction will also drive optical module and network switch upgrade demand, and companies such as Broadcom (AVGO) and Marvell (MRVL) continue to receive orders;
- Capital flows: Many foreign institutions now view data center infrastructure as the "shovel seller to AI's shovel sellers"; compared with highly volatile chips, infrastructure leaders offer more stable cash flows and valuation appeal.
Notably, Vertiv's earnings also confirm that AI computing investment is spreading from "chips to power." Goldman Sachs, a well-known investment bank, said in a report: "If AI is the electricity of the new era, then those providing power and cooling are the power plants of the new era. Companies like Vertiv will gain long-cycle endurance from the AI arms race."
Risk Warning: Super Cycle or Overinvestment?
However, the AI infrastructure sector is not without concerns. The market worries whether persistently high capital spending by Microsoft, Google, Meta and others will overdraw future demand. If AI commercialization returns fall short of expectations, capex could shrink sharply, hitting infrastructure companies more deeply than chip companies.
Additionally, Vertiv's forward P/E has exceeded 45 times, near historical highs. Investors should be wary of short-term valuation overshoot. Meanwhile, key components such as liquid cooling quick connectors and precision sensors still rely on a few suppliers, creating delivery delay risks.
Summary: A New Dimension for AI Computing Intelligence
For investors focused on AI computing in U.S. equities, Vertiv's earnings provide a new perspective: besides watching Nvidia's chips and Micron's HBM, the "hidden champions" of data center power and cooling deserve equal attention. As AI server density continues to rise, the growth certainty of liquid cooling and power infrastructure may surpass some chip segments.
Watch items going forward: Schneider Electric and Eaton's upcoming earnings, and the sustainability of Vertiv's liquid cooling orders next quarter. In addition, major cloud vendors' capital expenditure guidance for the second half of 2026 will be an important window to verify the supply-demand balance of AI computing power.
(This article is for industry analysis only and does not constitute investment advice. Investing involves risk; enter the market with caution.)
