Data centers, which are critical for handling vast amounts of digital information, consume significant amounts of energy, driven primarily by high-power processors. A recent Goldman Sachs report projects that by 2030, data centers will account for up to 8% of total U.S. energy consumption due to the growing demand for AI-driven services.
To address this escalating energy demand, Phaidra, a Seattle-based startup, is pioneering advanced artificial intelligence solutions designed to optimize data center power consumption and cooling systems. The company has recently secured $12 million in funding from Index Ventures, raising its total investment to $60.5 million. These funds will be used to advance research, development, and market expansion for Phaidra’s AI-powered virtual plant operator, which aims to enhance the efficiency, reliability, and sustainability of cooling systems in data centers and other critical infrastructures.
Meeting the Power Demands of Modern Data Centers
As the demand for high-power GPUs increases—driven by the needs of sophisticated AI algorithms—data centers face significant challenges in delivering and managing the necessary power. The relationship between computational capability and performance in AI systems is exponential, meaning that increases in computational power can lead to disproportionately large improvements in performance. This creates a competitive race to accumulate hardware and the energy required to power it.
Revolutionizing Cooling with AI
Cooling systems in data centers, which can consume between 20-40% of the total energy used, are a major focus for Phaidra’s AI technology. Their AI platform functions as a virtual plant operator, utilizing algorithms and data from thousands of sensors to optimize cooling parameters such as temperature, pressure, and flow rate. This dynamic control helps to efficiently remove excess heat while minimizing the energy used by cooling system components like pumps and chillers.
Phaidra’s AI employs reinforcement learning, a form of general intelligence that allows the system to improve its performance over time as it processes more data. This self-learning capability enables the AI to refine its algorithms continuously, enhancing efficiency autonomously without relying on external data sources.
Initially developed by Phaidra’s founding team at Google, this AI technology successfully reduced Google’s data center cooling energy consumption by up to 40%. Since spinning out as an independent company in 2019, Phaidra has reported that its initial customers have achieved energy savings of approximately 15% within the first year of implementation. These savings are anticipated to grow as the AI system becomes increasingly adept.
Expanding Applications
Beyond data centers, Phaidra’s AI technology has potential applications in other critical cooling systems, such as those used in pharmaceuticals and industrial settings. The platform’s autonomous and general AI capabilities also hold promise for broader industrial uses, including water conservation, equipment maintenance, and overall power utilization.
By leveraging AI to improve energy efficiency, Phaidra is setting a new standard for how data centers and other critical infrastructure can manage power consumption and cooling, contributing to more sustainable and cost-effective operations.