Jakarta, INTI - In today’s digital age, energy efficiency has become a primary focus in technology development. AI and accelerated computing, particularly through NVIDIA's continually enhanced twin machines, have demonstrated significant progress in reducing energy consumption across various sectors. This advancement is gaining widespread recognition due to its potential to enhance efficiency and reduce carbon footprints.
According to a recent report by Lisbon Council Research, a nonprofit organization focused on economic and social issues, even though data centers are predicted to contribute 4% of global energy consumption, AI can reduce the remaining energy consumption by up to 96%. The report highlights the supercomputer Leonardo from Italy, supported by nearly 14,000 NVIDIA GPUs, as a prime example of how this technology advances work in various fields, from car design to weather forecasting.
Accelerated Computing: Enhancing Energy Efficiency
Accelerated computing utilizes NVIDIA GPU parallel processing to perform more tasks in less time compared to servers using traditional CPUs designed to handle one task at a time. This method not only saves time but also significantly reduces energy consumption. For instance, shifting from CPU-based operations to GPU-accelerated systems can save over 40 terawatt-hours of energy annually, equivalent to the electricity needs of nearly 5 million homes in the United States.
User Experiences and Recent Innovations
Industries worldwide are already experiencing the benefits of AI and accelerated computing. In the financial sector, Murex, a Paris-based trading and risk management firm, tested the NVIDIA Grace Hopper Superchip. This CPU-GPU combination demonstrated a fourfold reduction in energy consumption and a sevenfold decrease in completion time compared to dedicated CPU systems. Pierre Spatz, Head of Quantitative Research at Murex, stated, “In risk calculations, Grace is not only the fastest processor but also far more energy-efficient, making green IT a reality in trading.”
In manufacturing, Taiwan-based Wistron used NVIDIA Omniverse to create a digital twin of its facility. Through the use of sensors and AI models, Wistron managed to improve overall facility energy efficiency by up to 10%, reducing electricity consumption by 120,000 kWh annually and cutting carbon emissions by 60,000 kilograms.
Positive Impacts Across Various Sectors
The RAPIDS accelerator for Apache Spark can reduce the carbon footprint of data analytics, a widely used machine learning method, by up to 80%, while providing an average speedup of five times and reducing computational costs by four times. Major companies, including Adobe, AT&T, and the IRS, utilize NVIDIA Spark accelerators for their large-scale data analytics.
In healthcare, Insilico Medicine used NVIDIA-supported AI to find and enter into phase 2 clinical trials a drug candidate for a rare respiratory disease. Using generative AI, Insilico achieved this milestone for a fraction of the traditional cost and time, which would otherwise exceed $400 million and take up to six years.
Addressing Challenges and Looking Ahead
Despite the significant benefits of AI and accelerated computing, some projections about the energy footprint of e-commerce and video streaming have proven to be overstated. Studies by the Information Technology and Innovation Foundation (ITIF) suggest that claims about AI’s environmental impact are often overly pessimistic and fail to account for the efficiency gains during AI model deployment.
AI holds immense potential to support sustainability efforts, including combating climate change. AI can enhance weather modeling accuracy, predict crop yields more precisely, and develop more accurate climate models. NVIDIA, along with various startups, is committed to leveraging AI to address environmental issues, including planning for the Earth-2 supercomputer, which will focus on climate science.
Conclusion
AI and accelerated computing offer innovative solutions to global energy challenges, enabling significant efficiency improvements across various sectors. As this technology continues to evolve, we can anticipate even greater contributions to energy sustainability and environmental protection.
For more information on how technology can advance innovation and sustainability, don’t miss the Indonesia Technology and Innovation 2024 (INTI 2024) Exhibition at JI-EXPO from August 12-14, 2024. INTI 2024 is the largest technology and innovation exhibition in Indonesia, featuring over 700 participants and four conference stages. For more details, visit INTI 2024.