Can AI find a way to reduce the broadcast industry’s energy consumption?

Energy and the modern data centre

The US Department of Energy’s Office of Energy Efficiency & Renewable Energy (EERE) cites the voracious energy appetites of the modern data centre, referring to them as the “most energy-intensive building types” with many times (between 10x to 50x) the energy needs of standard commercial buildings. The EERE points out that data centres make up nearly 2 per cent of all US electricity consumption. While vitally necessary to our economy, data centres most definitely present us with serious sustainability challenges.

The news, however, isn’t all bad—nobody’s suggesting that the power requirements aren’t worth it. However, through a combination of regulatory oversight, continuing technical advances in energy efficiencies, and increased corporate awareness of the problem space, data centres and the equipment they house will continue to occupy less physical space over time, all while consuming reduced amounts of energy and requiring much less human oversight and administration.

The better and more optimistic way to view data centre sustainability is to see that it’s a work in progress, with enormous upside for society, the economy and of course the environment, as we persist in overcoming the obvious and hidden challenges. Another hopeful sign is that data centres themselves might accelerate the solution to this entire sustainability problem.

Consuming energy to save energy

One of the newer denizens of the modern data centre is artificial intelligence, or AI, in the form of applications and digital services like ChatGPT, Adobe Firefly and Google Bard. Like technology predecessors such as enterprise-wide SaaS platforms and earlier enterprise database applications, AI is encountering explosive growth and rapid enterprise adoption. The rise of AI—and all the potential it brings with it—depends on the supporting infrastructures found in data centres around the globe, which means that AI becomes somewhat guilty by association in the entire sustainability question.

But the real question is this: is AI simply going to add more energy requirements for the modern data centre to meet, or can AI and all this new-fangled machine intelligence work some technological magic to help offset the demands, or even lower them beyond its own needs? And if the latter is true and AI is going to be part of the solution, in what ways can we expect this contribution to overall sustainability?

The smart money is already being placed on AI as being part of the overall solution. You only need to attend technology conferences such as the recently-held GTC AI conference in March to see the level of promise that AI holds in most areas of technology development, and that certainly goes for data centre sustainability.

To begin with, AI is assisting hardware and chip designers and fabricators not only in creating more powerful and robust products, but also in developing them with an eye toward reduced energy consumption and adaptive operations. What this ultimately means is that the next generation of servers and other equipment destined for data centre deployment will require far less power than current and past generations, be able to adapt automatically to operational conditions, and will have longer lifespans with reduced need for hardware refreshes.

AI also is capable of providing superior power monitoring, energy utilisation modeling, and operational optimisation which favours sustainable power sources in the course of daily activities within the data centre. With AI driving predictive modelling and energy forecasting needs, we can ensure that the most intelligent use of renewable energy sources is factored into overarching data centre operations, but in such a way that doesn’t risk reliability in pursuit of sustainability.

Emerging intellectual property points the way

One surefire way to see where technology is going involves keeping an eye on patent applications. Companies often reveal through their IP filings what’s on their mind, where their R&D budgets are moving them, and yes, what the future might have in store insofar as technology implementations within the data centre.

Intel recently filed a patent application for what could be the manifestation of a much more sustainable data centre. In the filing, the researchers detail the concept of a data centre applying AI and machine-learning techniques to understand the mix of renewable versus non-renewable energy available to the data centre, and then based on its advanced awareness of the machinery and power demands within the data centre, making more intelligent decisions about overall power consumption.

Patent applications such as these signal several different messages to us. First of all, we can clearly see that the biggest players in technology are dedicating more time and R&D energy to solving the sustainability problem within the data centre. This level of social consciousness should bring even the most cynical observer to an admission that technology can solve the very problems it helps to bring about—in this case, energy consumption.

Additionally, it showcases the fact that AI can augment (not replace) the skills of data centre operators and administrators, understanding far more about minute operational conditions and helping to make on-the-fly decisions with positive impacts on energy consumption. And lastly, ideas like these are necessary stepping stones to the broader adoption of AI and machine intelligence, which when applied correctly to the right problems, can help effect significant and lasting change.

A brighter and more sustainable future

While the thought of an all-powerful and menacing AI future works well in cinemas, we can clearly see that AI and machine intelligence, despite the obvious dependence on energy-hungry data centres, can help to transform the data centre landscape. Nearly everything within the data centre requires energy—AI can use that energy to monitor and solve the broader issue of sustainability surrounding it.