What will datacenters look like in 2030? Concerns around sustainability and the impact of AI will mean that datacenters at the end of the decade will be very different, built to be more sustainable, but also smarter and more powerful. The energy demands of datacenters are rising, with the International Energy Authority (IEA) saying they already account for up to 2% of global energy use, and this figure could double by 2026, equalling the electricity consumption of Japan. In the UK, the National Grid has predicted that the electricity demands of datacenters could potentially rise six-fold in just ten years.
What is behind this exponential surge? New technologies such as quantum computing and generative AI are predicted to drive a huge surge in demand for electricity. A preprint Carnegie Mellon study this year found that generating just one image with AI software uses as much energy as charging a smartphone. But AI is also predicted to drive a new wave of innovation which will help deal with the challenges of climate change, from limiting methane emissions from waste to designing ‘clean’ nuclear fusion reactors. One of the key challenges of this decade is to unleash the benefits of AI while managing its environmental impact.
Business leaders need to take a holistic view of the environmental impact of datacenters, including everything from the sources of energy used to how servers are shipped to the datacenters. Within the datacenter, there needs to be a focus on how servers are cooled and whether that energy can be reused in some manner, for instance to heat nearby buildings. To build datacenters fit for 2030 and beyond, business leaders need to understand the full impact of this technology, and then figure out how to implement it in the most sustainable way.
The AI revolution
With generative AI appearing in software ranging from email apps to internet search, the energy demands of the AI industry is going to continue rocketing, with one study in the journal Cell suggesting the power demand of AI alone could match the demand of the Netherlands by 2027. The compute power required to train AI doubles every six months, and Gartner® predicts that ‘by 2030, AI could consume up to 3.5% of the world’s electricity’. The IEA’s report suggests that adding generative AI to search (as companies including Google are rushing to do at present) could multiply the energy demands of internet searches by 10.
All of this requires the technology industry to design carefully for sustainability, not just at the chip level, but at the server level and datacenter level. It’s also worth bearing in mind that there is a flipside to this in terms of the environmental benefits of innovations sparked by new technologies. Emerging technology such as quantum computing will be more energy-efficient, and could also mean that problems are solved exponentially faster than classical computers. Both quantum and AI are expected to drive rapid innovation in everything from demand response in the electricity grid to photovoltaics to electricity generation technology. The ‘smart grids’ of the future will be powered by AI. In building decarbonisation, to take one example, Mckinsey estimates that AI can accelerate the process 100-fold compared to existing technologies.
Built for sustainability
The datacenter of the future will be designed from the bottom up with sustainability in mind. Technologies such as warm water cooling enable high performance with far less energy use, provided a host of power consumption benefits. For instance, in datacenters that use warm water-cooling, there is much less need for high-speed fans to dissipate heat. Air-cooling systems can often consume vast amounts of energy in themselves.
Furthermore, any wasted heat can be reused effectively because warm water-cooling systems produce heat waste at a temperature that is more easily reused for other purposes. Not only will this improve energy efficiency across an entire datacenter facility, but this energy can be successfully recycled in sustainable ways in the wider community. By 2030, recycling the heat from datacenter will become the norm, from heating nearby buildings and swimming pools to piping warm liquid under roads and walkways to melt ice.
The datacenters of the future will also be built around renewable energy, from renewable sources to solar panels on roofs, driving towards a future of carbon-neutral or even carbon-negative operations. Organisations will adopt ‘as a service’ approaches to AI to improve efficiency, and increasingly businesses will harness the power of AI to optimise electricity consumption in datacenters themselves. Just as AI will assist in demand response in the electricity grid, within the datacenter, algorithms can help with optimisation, predictive maintenance and energy management. This can reduce energy consumption by improving cooling efficiency, minimising waste and optimising resource allocation.
The ‘whole life’ view
When designing and building a datacenter fit for the future, it’s key for business leaders to take a view across the whole lifecycle of their datacenter and the servers that will work in it. Everything from how components are designed and manufactured to how they are shipped, deployed and disposed of at the end of their lives matters, and taking a holistic view is key to making real sustainability gains. Through this decade, asset recovery services (ARS) and recycling of computer equipment will grow in importance. Other markets in Europe have already implemented legislation to reduce the environmental footprint of digital technology. For example, France introduced a mandate that 20% of IT devices bought by organisations need to be refurbished, with a target of 40% by 2040.
Manufacturing products regionally to cut shipping miles will also be a key differentiator. The current shift towards ‘as-a-service’ approaches to everything from hardware to software will continue, with business leaders focusing on avoiding overprovisioning to cut carbon emissions.
Sustainable datacenters
This decade will see substantial progress towards smarter, more sustainable datacenters, even as AI drives a new surge in energy demands. On the surface, the datacenter of 2030 might look similar to the ones we have today, but careful thought will have reshaped its impact on the wider world. Everything from how servers are cooled to how resources are allocated will be done more intelligently, and renewable energy will enable a new kind of carbon neutral datacenter. These datacenters will be integrated into their local communities, providing warmth, while also providing the computing muscle for AI to deal with big global challenges, such as climate change.
You might also like
- We've rounded up the best Large Language Models (LLMs)
- And here are the best LLMs for coding
- Why the future of AI is nuclear
