Published on augustus 15, 2024, 12:26 am
Image source: Fox News
```json [ "TLDR: Recent study shows AI energy consumption may surpass Bitcoin mining due to higher revenue generation, leading some miners to transition to AI technologies. Generative AI models have substantial energy needs, potentially exceeding Bitcoin by 2024. Challenges include the impact on electricity and water usage, reliance on continuous power supply, and potential use of less eco-friendly energy sources. Despite profitability, transitioning between technologies is complex, with concerns about escalating energy demands in pursuit of larger AI models.", "AI data centers require proximity to urban areas for low latency responses compared to location-flexible Bitcoin mining operations that can utilize surplus renewable energy sources. Sustainable practices and technological innovations are seen as essential for managing the increasing energy demands of both AI and cryptocurrency mining." ] ```
A recent study suggests that the energy consumption of artificial intelligence (AI) systems might already exceed that of Bitcoin mining operations. The research conducted by the Bitcoin Policy Institute indicates that AI is becoming a formidable rival to Bitcoin in terms of electricity usage and hardware demand.
AI companies, benefiting from substantial financial resources, are capable of outbidding Bitcoin miners for electricity. This competition arises because AI can generate significantly higher revenue, up to 25 times more per kilowatt hour than Bitcoin mining. Consequently, some miners are integrating AI processing into their data centers or even transitioning entirely from Bitcoin to focus on AI technologies.
Although the AI industry is still in its early stages, the energy requirements for generative AI models are substantial. For instance, a single ChatGPT query consumes nearly 10 times more energy than a typical Google search. Moreover, creating an AI-generated image can consume as much power as fully charging a smartphone.
The growth in generative AI adoption suggests that by 2024, AI could consume 169 terawatt hours yearly, surpassing Bitcoin mining’s estimated 160 terawatt hours. The demand for AI computing not only impacts electricity usage but also water consumption for cooling data center equipment.
Unlike Bitcoin mining that offers flexibility to scale operations based on electricity availability, AI models require continuous power supply for optimal performance. This necessity may lead to the utilization of less environmentally friendly energy sources like fossil fuels in maintaining uptime.
Despite the profitability of AI compute compared to Bitcoin mining, transitioning from one technology to another poses challenges due to differences in hardware specifications. While advancements in hardware and algorithms could enhance energy efficiency for AI models, there remains a concern about the escalating energy demands driven by the pursuit of larger models.
In contrast to location-agnostic Bitcoin mining, where operations can adapt to surplus renewable energy sources, AI data centers necessitate proximity to urban areas for low latency responses. Nonetheless, proponents argue that sustainable practices and technological innovations are crucial for addressing the increasing energy demands posed by both emerging technologies like AI and established ones like cryptocurrency mining.
Original article posted by Fox News
