The surge in AI technology is leading to increased electricity consumption.
A recent investigation in May 2024 by Goldman Sachs forecasts that by 2030, data centers could consume 8% of the total U.S. power supply, a significant rise from 3% in 2022, due to the expansion of cloud services for AI systems. This projected trend suggests that to accommodate both the modern and emerging AI data centers, approximately $50 billion will be needed for additional power generation capacity in the U.S.
Such explosive growth is not without its challenges. For instance, Meta’s recent commencement of a large data center project in Kansas has prompted Evergy, a local power provider, to postpone the shutdown of a coal plant for five years, as reported by Kansas Reflector. The extensive energy and water usage by these data centers could lead to higher utilities expenses, disproportionately affecting those with lower income.
Despite these issues seeming daunting, the founders of Phaidra, Jim Gao, Katie Hoffman, and Vedavyas Panneershelvam, see a viable solution in making existing data centers more energy-efficient through retrofits.
Their venture, founded in 2019, is at the forefront of developing AI-driven controls for data centers, pharmaceutical, and commercial buildings to optimize energy use. Phaidra’s technology employs a network of sensors within these facilities to enable real-time, energy-saving decisions, particularly in cooling systems which are among the largest energy drains, consuming up to 40% of a data center’s total power according to a McKinsey report.
Gao, who previously spearheaded Google’s DeepMind Energy for commercializing climate change mitigation technologies, emphasizes the innovative approach of Phaidra. After parting ways with DeepMind Energy, Gao, alongside Panneershelvam, decided to leverage their expertise in AI and energy optimization in other sectors, including data centers.
A special focus of Phaidra is on merging physical system knowledge with AI-driven models to enhance facility operations, based on extensive sensor data. This strategy enables the creation of adaptive models that continuously refine their energy-saving tactics.
While competitors exist in the quest to reduce data center energy consumption, Gao believes Phaidra’s main challenge is the conventional methods of energy management. Traditional techniques often involve infrequent updates and lack the adaptability that AI provides, making Phaidra a more dynamic solution.
Interestingly, Phaidra’s initial client base extended beyond data centers, including major pharmaceutical firms like Merck. Yet, the demand within data center operations has significantly grown, propelled by the burgeoning interest in AI.
Recently, Phaidra’s potential for impact was recognized by being selected as a finalist in the Amazon Sustainability Accelerator, a move that could align well with its ambitions for expansion, especially in regions facing high energy costs.
The company, based in Seattle and boasting a team nearing 100 employees, secured $12 million in a financing round led by Index Ventures, bringing their total funding to over $60 million. These resources are earmarked for further research, development, and scaling their market presence.
As Phaidra sets its sights on expanding its team and operations, the path forward looks promising for this pioneer in making AI and data centers more sustainable and efficient.
Compiled by Techarena.au.
Fanpage: TechArena.au
Watch more about AI – Artificial Intelligence
