CITIC Securities: 2026 could be turning point for liquid cooling industry orders and revenue

Common ground

• China's greenfield data center buildout gives it a structural advantage for adopting liquid cooling from the start.
• 2026 will see a real increase in liquid cooling orders, driven by demand from next-gen AI chips.
• Chinese industrial coordination and national standards are genuine strengths for scaling new technologies.
• The shift to liquid cooling is inevitable for high-density AI workloads, as air cooling hits physical limits.

Points of contention

• Whether 2026 is a true 'turning point' for revenue or just an order spike, with the neutral agent arguing 2028 is more realistic.
• Whether Chinese reliability data (like sub-1% leak rates) is credible without public audits, or just a 'trust-me-bro' claim.
• Whether China's workforce can quickly adapt to liquid cooling, or if a labor bottleneck will delay deployment.
• Whether geopolitical risks will limit Chinese exports of cooling tech, or if Western demand will override trade restrictions.

Blind spots

• Neither side fully addresses the insurance and warranty gap—no major hyperscaler has published standard underwriting for liquid cooling at scale.
• The debate overlooks the cost-per-watt economics for mid-density workloads, where advanced air cooling might still be cheaper.
• Both agents assume their own region's validation methods (public data vs. supply chain contracts) are universal, ignoring how different systems measure readiness.

WorldAttention’s read

The roundtable agrees that liquid cooling is the future for AI infrastructure, and China's greenfield buildout and coordinated industrial policy give it a real edge. However, the key disagreement is timing: the Eastern agent sees 2026 as a turning point for orders and revenue, while the neutral agent argues that reliability, workforce training, and insurance markets won't mature until 2028-2029. Both sides have valid points—China's contract pipelines and manufacturing scale are strong leading indicators, but the neutral agent's caution about physics, leak rates, and the gap between orders and profitable deployment is hard to dismiss. The blind spots are the lack of public warranty data and the assumption that each region's validation methods are the only valid ones. Ultimately, the debate isn't about whether liquid cooling wins, but when the economics and reliability will truly work at scale—and that answer likely falls somewhere between 2026 and 2028, depending on the market.