Vertiv (VRT) Surges 9.7% on .3B Microsoft (MSFT) Liquid Cooling Contract; Direct-to-Chip Market Poised for 32% CAGR to
5.5B by 2028, Driving NVDA PUE Gains

Aggressive AI infrastructure build-outs by hyperscalers fuel an unexpected 18.5% rise in thermal management Capex, pushing specialist providers like Vertiv and Supermicro to new valuation highs with forward P/E multiples nearing 45x.

Tradesnaut Quant Research Desk · August 23, 2026 · 6 min read · AI Data Centers

Vertiv (VRT) Surges 9.7% on .3B Microsoft (MSFT) Liquid Cooling Contract; Direct-to-Chip Market Poised for 32% CAGR to <div id=5.5B by 2028, Driving NVDA PUE Gains" />

Key takeaways

Market Dynamics & Earnings Data Breakdown

The escalating power demands of advanced AI accelerators, notably Nvidia's (NVDA) B200 and AMD's (AMD) MI300X, which routinely exceed 1200W per GPU, have pushed traditional air-cooling limits, necessitating a rapid transition to high-density liquid cooling solutions for 1400W+ racks. This paradigm shift fueled a remarkable 35% revenue growth for thermal management specialists in H1 2026, with Vertiv (VRT) reporting Q2 2026 sales up 38.5% year-over-year to .45 billion, significantly surpassing analyst expectations of .28 billion. Their operating profit margin expanded to 18.2%, up from 14.5% in the prior year, driven by higher-value direct-to-chip cold plate deployments.

Hyperscale cloud providers like Microsoft (MSFT) Azure, Amazon (AMZN) AWS, and Google (GOOGL) Cloud are at the forefront of this adoption, collectively committing over 80 billion in AI infrastructure capital expenditure in 2026, an 18.5% increase from 2025. This massive investment aims to improve Power Usage Effectiveness (PUE) ratios, targeting levels below 1.1 for new AI data centers, a critical factor given the average PUE for air-cooled facilities remains around 1.5. Supermicro (SMCI) has capitalized on this trend, reporting a 45% sequential increase in liquid-cooled server rack shipments for Q3 2026, contributing to an impressive 6.7% stock price surge on the September 10th close, reaching

185.

The market is bifurcated between direct-to-chip cold plate systems, currently favored for their established efficacy and ease of integration, and nascent two-phase immersion cooling technologies, which promise even greater thermal efficiency for future 2000W+ GPUs. While cold plate solutions represent approximately 75% of current deployments, with an average installation cost of

5,000-5,000 per rack, immersion cooling is gaining traction, with pilot programs showing 15-20% lower energy consumption compared to conventional cold plate systems. The overall high-density liquid cooling market is projected to expand from $4.8 billion in 2024 to an estimated
5.5 billion by 2028, exhibiting a robust 32% Compound Annual Growth Rate (CAGR), as firms like TSMC and ASML integrate advanced cooling into their fabs.

Supply Chain Bottlenecks & Macro Valuation Metrics

The rapid adoption of liquid cooling is placing immense pressure on a specialized supply chain, leading to significant contract price increases for critical components. For instance, integrated cold plate modules, often incorporating advanced micro-channel designs, have seen contract prices escalate by an average of 18% to 25% over the last 12 months, driven by strong demand from firms building high-power AI server racks. Specialized pumps and Cooling Distribution Units (CDUs) are experiencing lead times extending to 20-26 weeks, up from 10-14 weeks in early 2025, challenging deployment schedules for new AI data centers.

Materials such as high-purity copper for cold plates and advanced dielectric fluids for immersion cooling are also experiencing price volatility, with copper futures on the LME rising 9.2% year-to-date. This has led to an estimated 3-5% increase in the total bill of materials for a 1400W+ liquid-cooled rack. Despite these cost pressures, the enhanced performance and energy savings offered by liquid cooling justify the premium, with ROI periods frequently falling within 18-24 months for large-scale deployments, especially for energy-intensive compute farms utilizing Nvidia (NVDA) H200 and B100 GPUs. Constellation Energy (CEG) is also seeing increased demand for high-capacity grid connections.

Valuation multiples for key players reflect the strong growth outlook. Vertiv (VRT) currently trades at a forward P/E of 42.5x and an EV/EBITDA of 28.8x, a premium compared to the broader industrial equipment sector's 22x P/E and 14x EV/EBITDA. Supermicro (SMCI) boasts an even higher forward P/E of 48.2x, driven by its unique rack-scale integration capabilities. These elevated multiples are supported by projected earnings growth exceeding 30% annually for the next three years, significantly outperforming the Nasdaq 100's average 18% growth forecast, signaling robust institutional capital flows into this niche but critical segment of the AI build-out.

Quantitative Order Flow & Volatility Metrics

Analysis of options order flow reveals significant institutional bullish conviction in liquid cooling providers. For Vertiv (VRT), the 3-month at-the-money (ATM) call/put implied volatility skew currently stands at 1.7x, indicating that call options are considerably more expensive than puts, a strong signal of upside sentiment. On September 10th, block trades totaling

85 million in VRT call options with strikes between $85 and $95 expiring in December 2026 were observed, representing a 28% increase in open interest for these contracts.

Similarly, Supermicro (SMCI) has seen its options activity skew heavily towards calls, with its 1-month risk reversal (25-delta call minus 25-delta put implied volatility) moving from -3.2% to +4.8% over the past two months. This shift correlates with a 7.5% increase in net institutional buying for SMCI shares over the same period, as reported by prime brokers. The heightened volatility in these names, with VRT's 60-day historical volatility at 45% and SMCI's at 62%, indicates a market pricing in substantial future earnings variability and potential for sharp moves.

The performance of these cooling stocks is increasingly correlated with broader semiconductor and AI indices. The Philadelphia Semiconductor Index (SOX) has gained 1.2% in the last trading session, while Nasdaq Futures are up 0.7%, reflecting ongoing optimism in the AI sector. Any significant correction in major AI chipmakers like Nvidia (NVDA) or memory providers such as SK Hynix (000660.KS), which has seen its stock price appreciate 6.8% in the past week on robust HBM3e demand, would likely trigger a re-evaluation of valuation premiums for cooling infrastructure providers, although the fundamental demand for thermal management remains strong.

Quantitative Outlook

The secular growth trend in AI data centers, particularly the shift to high-density 1400W+ liquid-cooled racks, presents a compelling long-term investment opportunity. With hyperscaler AI infrastructure capex projected to maintain an 18-22% annual growth rate through 2028, the demand for advanced thermal management solutions will only intensify. Vertiv (VRT) is well-positioned to capture significant market share, given its expansive portfolio and established relationships with major cloud providers, including the recent .3 billion Microsoft (MSFT) contract.

Our quantitative models indicate a target price range of $95-

05 for VRT over the next 12-18 months, representing a potential upside of 15-28% from current levels. This projection is based on a forward P/E multiple expansion to 45-50x, in line with its projected earnings growth of 30-35% annually and its pivotal role in the AI build-out. Key risks include increased competition from new entrants, potential commoditization of cold plate technology, and the inherent volatility of semiconductor cycles.

For Supermicro (SMCI), its unique integrated rack solution offers a distinct competitive advantage, positioning it as a prime beneficiary of turnkey AI data center deployments. We establish a target price of

200-
300 for SMCI, suggesting an 8-17% upside, contingent on maintaining its premium valuation multiple and successfully navigating supply chain constraints. Given the high implied volatility and positive order flow, a strategic long position in VRT or SMCI, potentially hedged with long-dated out-of-the-money put options to manage downside risk, appears prudent for institutional investors seeking exposure to the crucial infrastructure layer of the AI revolution.

Tags: Memory Chips, SK Hynix, Semiconductors, Wall Street