Tech News
Back to home
Technology

In a first, Google has released data on how much energy an AI prompt uses

Source

MIT Technology Review

Published

TL;DR

AI Generated

Google has released a technical report revealing that the median energy consumption of its Gemini AI apps for a single query is 0.24 watt-hours, equivalent to running a microwave for one second. The report offers insights into water consumption and carbon emissions associated with AI prompts. The study highlights the energy demand breakdown, showing that AI chips account for 58% of the total electricity demand, with other infrastructure components also playing significant roles. Google's transparency in sharing this data is seen as a significant step in understanding AI energy usage, with experts praising the comprehensive analysis provided. The company's efforts to optimize energy efficiency in AI models have led to a significant reduction in energy consumption over time.

Read Full Article

Similar Articles

Inventor showcases 3D printer filament dryer that mines Bitcoins and dries filament with waste heat, capable of 6 TH/s at 140W — joins Bitcoin-mining 3D printer in hobbyist-focused miner lineup

Inventor showcases 3D printer filament dryer that mines Bitcoins and dries filament with waste heat, capable of 6 TH/s at 140W — joins Bitcoin-mining 3D printer in hobbyist-focused miner lineup

An inventor has created a filament dryer that not only dries filament for 3D printing but also mines Bitcoins, utilizing waste heat from the process. This new prototype is capable of 6 TH/s at 140W and is part of a lineup of hobbyist-focused mining devices. The filament dryer helps prevent common printing issues caused by moisture absorption and offers a proactive solution for maintaining filament quality. It is likely that this new device shares underlying technology with the inventor's previous 3D printer that also doubled as a Bitcoin miner, hinting at further scalability and innovation in this unique niche market.

Tom's Hardware•
Nvidia-backed trial shows AI data centers can flexibly adjust power use in near real time, with global implications for energy consumption — suggests hyperscalers can reduce consumption as necessary, ensuring grid isn’t overloaded during peak demand

Nvidia-backed trial shows AI data centers can flexibly adjust power use in near real time, with global implications for energy consumption — suggests hyperscalers can reduce consumption as necessary, ensuring grid isn’t overloaded during peak demand

A recent U.K. study involving Nvidia and other partners demonstrated that AI data centers can dynamically adjust power consumption, potentially reducing strain on the grid during peak demand. The trial showed that hyperscalers could swiftly modify power usage, tapping into renewable energy surplus during low demand periods. While this flexibility may not align with the need for maximum power utilization by hyperscalers, it could expedite data center deployment by relieving grid pressure. If data centers and grid operators collaborate on such adaptive systems, it could optimize power usage and facilitate quicker infrastructure connectivity.

Tom's Hardware•
Enthusiast runs desktop PC off 56 AA batteries — Intel computer lasts less than 5 minutes while playing Minesweeper

Enthusiast runs desktop PC off 56 AA batteries — Intel computer lasts less than 5 minutes while playing Minesweeper

A YouTube creator experimented with running a desktop PC off 56 AA batteries, replacing the power supply with an ATX plug. Despite various attempts, including using alkaline batteries and wiring tweaks, the system only lasted a few seconds before shutting down. The creator managed to play Minesweeper for 4 minutes and 35 seconds on battery power alone but aimed to beat the game entirely on AA power. Adding a GPU allowed the PC to run for around nine seconds, showcasing advancements in battery technology. Suggestions were made to try the experiment with laptop components for lower power draws.

Tom's Hardware•
TSMC reduces peak power consumption of EUV tools by 44% — company to save 190 million kilowatt-hours of electricity by 2030

TSMC reduces peak power consumption of EUV tools by 44% — company to save 190 million kilowatt-hours of electricity by 2030

TSMC has launched the EUV Dynamic Energy Saving Program to reduce the power consumption of its EUV lithography tools, aiming to save 190 million kilowatt-hours of electricity and cut carbon emissions by 101 kilotons by 2030. The program, which involves adaptive power scaling based on real-time operational status, has already reduced peak power consumption by 44%. TSMC plans to implement this program across all its EUV tools globally and in future fab installations. Despite the significant energy savings, it represents only a small portion of TSMC's overall power consumption, with most power being consumed by supporting systems.

Tom's Hardware•

We use cookies

We use cookies to ensure you get the best experience on our website. For more information on how we use cookies, please see our cookie policy.