As artificial intelligence (AI) rapidly evolves and we become more dependent on its capabilities, its impact on data centres is becoming increasingly significant.
How is the surge in AI-driven applications affecting data centre operations?
Will existing facilities be able to support this growth?
In this blog, Future-tech’s Head of Technical Due Diligence Mark Acton delves into these questions, exploring how AI is reshaping data centre requirements and energy usage.
AI’s Rising Power Needs and Data Centre Power Impact
With all the attention gained by Chat GPT etc. performing impressive creative tasks, it is easy to overlook the significant amounts of power needed to provide to these services. Some researchers closely involved in the development of these platforms are saying that use of the current Large Language Modules is unsustainable in terms of energy consumption.
Where do these platforms sit? In data centres of course!
The power needed to support LLMs is already being recognised and is clearly adding to data centre power consumption. This is inevitably going to be a growing problem for the current models, some of which rely on existing ‘unused’ processing capacity and are therefore currently largely hidden in terms of data centre power usage increases.
As power consumption rises, this will inevitably result in increasing calls from poorly informed media and governments to curb data centre power usage while failing to understand where this demand comes from – All of us using Chat GPT etc!
Additionally, the large power demands of the new GPU Chipsets from Nvidia designed for AI functions are going to be increasingly difficult to host in traditional data centres. A single chipset can consume 1 kW with some individual servers reportedly consuming up to 17 kW. This will produce both problems and opportunities for data centre operators, ironically some of which may be solved using AI!
Artificial General Intelligence (AGI) is very different to Generative AI. AGI emulates human intelligence and remains theoretical with some even doubting that it will ever be truly possible, others saying it may be possible by the end of the decade. Generative AI however is with us now and is very good at solving specific problems or dealing with very specific tasks by offering a statistically valid representation based on available content, rather than being truly ‘intelligent’.
Artificial Super Intelligence (ASI) is intelligence above and beyond human and although very theoretical is where things get really interesting and words of caution become meaningful. In this case the word ‘Artificial’ may even have to be dropped. Generative AI is not going to take over the planet – ASI might!
It is frequently noted though that AI is probably essential to solving climate change and energy usage issues in data centres, and far more so we should not be too quick to dismiss it. It is becoming increasingly apparent that we need intelligent tools to manage the increasing power density and cooling demands of AI workloads. Managing the complexity of very power dense data centres with millions of data points is beyond the capacity of humans to manage, so we increasingly need tools in data centres to manage AI workloads reliably and efficiently.
One thing is certain, AI will inevitably follow Amara’s Law, the initial promise may be inflated, but the long-term impact is likely to be very much underestimated.
…
Mark’s expert insight is part of Inside Networks Magazines’ ‘Question Time’ segment. To read more answers by industry leaders, check out the September 2023 Issue from page 19.
…
Future-proofing Data Centres with Future-tech
At Future-tech, we specialise in designing and optimising data centres to handle these increasing demands. With our expertise in energy-efficient infrastructure and cutting-edge technologies, we can help your facility adapt to the evolving requirements of AI, ensuring resilience, sustainability, and long-term success.
Contact our expert team today to learn how we can future-proof your data centre project.