DARPA Funds Project to Track AI Energy Usage Across Workflows
The U.S. Defense Advanced Research Projects Agency (DARPA) is investing $1.83 million over two years to support the E-CALC initiative, aimed at accurately measuring energy consumption of artificial intelligence (AI) throughout every stage of its workflow. The project involves researchers from Michigan Engineering, the University of Illinois Urbana-Champaign, and BAE Systems.
AI’s Rising Energy Demand and Its Impact
The rapid adoption of AI technologies is dramatically increasing power demands globally. Current projections by the International Energy Agency anticipate that data centers’ electricity consumption will more than double by 2030 compared to 2024 levels, escalating from approximately 420 terawatt-hours (TWh). This surge stems largely from the widespread integration of machine learning into sectors ranging from energy grids to consumer electronics.
Although AI holds promise to improve efficiency and reduce energy consumption in various fields by optimizing processes and discovering new materials, its own energy footprint remains enormous and poorly understood. A critical insight from Michigan Engineering highlights that inefficiencies in aligning AI software with hardware components cause up to 30% of wasted electricity during AI model training.
Why Monitoring AI Energy Use Matters for Pakistan
Pakistan, like many developing nations, is rapidly adopting AI-driven technologies across industries such as telecommunications, finance, and energy management. However, with its ongoing challenges in energy supply and infrastructure, escalating and unmeasured energy demands from AI could stress the country’s power grid and increase reliance on fossil fuels if left unchecked.
By developing tools such as E-CALC to precisely quantify AI’s energy consumption, stakeholders can identify inefficiencies and adopt greener, more cost-effective AI solutions. This aligns with Pakistan’s broader objectives to expand clean energy capacity and meet international climate commitments, while ensuring sustainable technological progress.
Advancing AI Energy Efficiency Through Innovation
The E-CALC project addresses a crucial challenge: the absence of a standard method to measure AI’s energy use at every stage—training, validation, and deployment. With better data, researchers can design AI architectures and workflows that reduce energy waste and environmental impact.
According to Michigan Engineering, “misalignments between software and hardware waste massive amounts of electricity—up to 30% during AI model training alone.” Improving this alignment can significantly lower the carbon footprint of expanding AI applications worldwide.
Ultimately, initiatives like this underscore the importance of making AI itself more energy-efficient, balancing its transformative potential with sustainability goals.
Related Stories
- UK Greenlights Jackdaw Gas Field, Stirring Energy Debate
- China May Deploy Space Laser Weapons by 2030: US Intel
- Change in petrol pricing formula causes serious set back to refineries: CEO ARL
- Diesel Prices Surge Past Rs400, Petrol Rates Hold Steady
- Bolsonaro Defeat Challenges US Plans for Brazil Rare Earths
