Ampere’s Chief Product Officer, Jeff Wittich, shares his thoughts on effectively managing data center growth while prioritizing sustainability.
At the close of 2023, the global tally of data centers neared 11,000, with the United States, Germany, the UK, and China leading the charge in locations. Recent insights from Cushman & Wakefield highlighted the largest data center markets, gauged by electricity capacity and consumption. The Northern Virginia data center market reigns as the world’s largest, handling over one-third of global online traffic. In the past year alone, data centers devoured 7.4 Gigawatts of power, a 55% increase from the previous year.
The emergence of disruptive technologies, particularly AI, has triggered an exponential surge in data traffic, propelling the demand for additional data centers and power. According to a recent report, the International Energy Agency (IEA) expects to see a “rapid pace” of growth for data center electricity consumption in the US over the next couple of years, rising from roughly 4% of US demand in 2022 to 6% by 2026.
The growing interest in eco-friendly data center solutions is driving the growth of the cooling markets. Projections indicate that spending on data center cooling technologies will increase from $12.7 billion in 2023 to $29.6 billion by 2030. However, cooling operations require substantial volumes of water, posing a burgeoning challenge to communities. For instance, Iowa, which has emerged as a real estate hub for tech giants, received a stark warning about declining groundwater levels due to the massive water demands of data centers.
Meanwhile, nations across the globe are exploring unconventional solutions. China made waves last year by deploying the “world’s first commercial underwater data center,” leveraging the ocean as a natural cooling mechanism. This innovative approach is projected to enhance energy efficiency by 40% to 60%, but accessibility and water temperature concerns linger.
As advancements increase, it has become evident that AI is no longer a mere differentiator but imperative for business expansion. Nearly 77% of devices today use AI technology, and 84% of C-level executives believe they must adopt and leverage the technology to drive business objectives. Tech luminaries have weighed in, offering strategies to mitigate the environmental repercussions of data center expansion fueled by AI. Sam Altman, the head of OpenAI, highlighted nuclear fusion as a prospective solution. However, the industry swiftly countered, arguing that the rapid rate of AI innovation requires more near-term solutions. To keep pace, we need a resolution now.
Fortunately, a straightforward solution lies at the heart of the data center: the CPU. While the industry clamors for more GPU power, the reality is that most workloads do not necessitate a GPU – 85%, to be exact. Ampere’s GPU-Free Platforms, featuring Cloud Native Processors and Optimized AI Frameworks, maximize performance, energy efficiency, and affordability for AI inferencing workloads. When running OpenAI’s Whisper model, Ampere CPUs offer up to 2.9 times the performance of a number of Nvidia’s GPU-based solutions. Depending on the comparative GPU solution, they can consume substantially less power per inference.
Also Read: Nvidia Unveils ‘Blackwell’ Chip, AI Robots at GTC 24
The evolving data center landscape faces a critical juncture, balancing growth, efficiency, and environmental concerns as data demands surge, propelled by disruptive technologies like AI, and sustainable responsibility intensifies. While eco-friendly solutions and innovative approaches offer promise, challenges persist. Optimizing foundational components like CPUs is pivotal. Learn how Ampere‘s cloud-native processors deliver the only sustainable computing platform with predictable performance, scalability, and power efficiency.
