Industry expert discusses liquid cooling and sustainability solutions for scaling AI datacenters.

Dr. Kelley Mullick, founder and CEO of Avayla, a consulting firm specializing in AI data center design, liquid cooling solutions, and cloud infrastructure optimization, has emerged as a leading expert in AI infrastructure with nearly two decades of experience spanning hyperscaler enterprise and edge industries. As artificial intelligence penetrates all industries, the infrastructure supporting it faces an urgent challenge: keeping up with the immense power and cooling demands of AI-driven workloads. Mullick emphasizes the fundamental changes required, stating: "For AI to function at scale, we need a fundamental shift in how we build data centers. Liquid cooling is part of that equation, but it's just one piece. We're talking about a complete overhaul of infrastructure to support the next era of computing."

The technical imperative for infrastructure transformation stems from the exponential growth in AI model complexity and size. The latest GPUs powering these models demand thermal design power levels far beyond what traditional air cooling can support. "Right now, most of the world's data centers still rely on air cooling," explains Mullick. "But AI has pushed thermal requirements beyond what air can handle. We're seeing companies scramble to adapt, but without the right infrastructure in place, AI clusters risk inefficiency, overheating, and instability." Scaling AI further requires pooled compute resources across multiple racks, unlike traditional workloads where each rack operates independently, necessitating a complete reimagining of data center cooling and networking.

Throughout her career, Mullick has been at the forefront of liquid cooling adoption in data centers, having led warranty strategies for immersive cooling and developed industry-first total cost of ownership models demonstrating the economic benefits of liquid cooling. She anticipates a hybrid future of cooling solutions: "Many in the industry assume direct-to-chip cooling is the only answer. While it's a mature technology, immersion cooling and two-phase technologies are also growing. We're going to see a hybrid future, where multiple cooling solutions work together to meet AI's escalating demands." Beyond technical advantages, liquid cooling delivers compelling economic and environmental benefits. Mullick notes that "switching to liquid cooling can cut data center operating costs by up to forty percent," a significant advantage as AI-driven power consumption forces companies to reconsider energy strategies. Additionally, liquid cooling drastically reduces water consumption, crucial in water-scarce regions. "In places like Arizona or the Middle East, where water is precious, reducing reliance on traditional cooling methods seems like the most logical aspect," she observes.

Beyond her technical achievements, Mullick is recognized as a prominent advocate for diversity in technology and has been instrumental in advancing liquid cooling standards. She has been named among the Top 50 Women Leaders in Technology (2024, 2023), Mission Critical Magazine's Top 25 Women in Technology (2024), and Top Women Leaders in Technology (2023) by Women We Admire. As an inventor, she has been part of a patented software system using AI training to prioritize and resolve customer complaints, resulting in a new revenue stream of approximately $60 million. With several patents and groundbreaking work that has driven significant reductions in tail latency and boosted throughput in cloud infrastructures, Mullick serves as chair of the Industry Liaison Team within the Open Compute Project, actively working with hyperscalers and industry leaders to set standards for next-generation data centers.

The urgency of infrastructure evolution cannot be overstated. "The biggest challenge right now is scaling the infrastructure fast enough to meet AI's demand," Mullick says. "We're seeing hyperscalers struggle with stability issues at scale. AI infrastructure has to evolve beyond today's assumptions." Her upcoming speaking engagements at the Open Compute Project's Dublin event will further solidify her role as a thought leader in AI-driven data center evolution. As AI reshapes industries, the companies that invest in next-generation AI infrastructure today will be positioned to lead tomorrow.