Chris Sharp, CTO, Digital Realty
14 February 2023
Predicting the future is hard. Predicting the future of technology and data centre technology is even harder. Technology – by design – evolves so rapidly, often in response to our ever-changing needs as a species. Take the global health pandemic, for example. Under normal circumstances, producing a vaccine can take anything from ten to 15 years. The development of the COVID-19 vaccine took under a year, highlighting the power of technology and innovation.
2022 was an interesting year for advancements in data center technology. Despite being a year of both recovery and rediscovery, technological innovation remained strong; whether it was improvements in our understanding of AI and its relationship with consciousness, or breakthroughs in gene-editing technology. Underpinning most of this technological innovation is the data centre industry: the central nervous system of the digital economy. And with 2022 ending, it feels like a good time to look at what 2023 – and beyond – could have in store for technology and data centres, as well as how macro trends – such as geopolitical conflict and ongoing concerns for energy security – could affect it.
Data centres will wave ‘goodbye’ to fossil fuels in favour of renewable alternatives
Renewable energy in data centres is something that the data centre industry has been moving towards for many years now. However, concern over energy security – an issue that really came to the fore this year because of the ongoing conflict in Eastern Europe – has sharply prompted the industry to closely examine its partial reliance on unrenewable energy sources, like diesel or gas.
There are lots of thriving data centre hubs around the world, like London, Singapore, and Virginia, that still have so much growing to do but are constrained by energy, both in terms of access and supply. That’s not to say that we don’t plan how much capacity we need in advance. When we look to expand, there are several factors we take into consideration. Planning for future capacity needs is one of them, but we also look at, for example, aligning with the grid’s capacity and whether we can get long-term supply agreements in place. However, for these hubs to reach their full potential and keep up with the demand for digital services globally, the industry needs to start looking at alternative ways to power its facilities, which ultimately means adopting a hybrid model and having the ability to go off grid and become ‘decentralised’.
It sounds unrealistic now, but if you examine history, it’s not. For instance, it wouldn’t have been uncommon for a factory in the 1920s to have been powered by a single steam engine located right outside, essentially acting as a ‘microgrid’ for the factory. Just imagine that but on a much broader scale, and instead of steam engines, think about renewable power, such as solar, wind, nuclear, and hydropower. We already do this to some extent now, with more than 900 megawatts of solar and wind energy under contracts in the US alone. But it won’t be long until fossil fuels take a back seat permanently, and data centers become self-sustaining in terms of power, able to switch on and off the grid at the flick of a switch.
The explosive adoption of data-driven technologies will lead to the mainstream emergence of innovative cooling methods
The use of data-heavy technologies like AI and IoT continues to ramp up across the globe. In fact, since 2021, 86% of CEOs reported that AI is considered a ‘mainstream technology’, and 91.5% of leading businesses invest in AI on an ongoing basis. Whether its live sentiment analysis tracking in the advertising industry or the use of complex algorithms by social media companies, businesses are seeing tangible value from the use of AI. And if businesses keep seeing value, usage is only going to increase, which means more intensity and more distribution geographically.
The thing is, AI doesn’t run on thin air. It’s a demanding technology, requiring a lot of power to run, and incredibly high-power density racks to function optimally. And, if high-power density racks are needed, so is high-power effective cooling. Our data centres are modular, which means components can be swapped out for other parts which might be better suited to a specific scenario, allowing us to evolve with increasing power densities.
Now, historically, the most effective way of cooling a data centre has been through hot and cold isle containment. Or, if you’re lucky, you might be able to make use of a local water source, like a river or dock, to cool your data centre naturally, like we do in Marseille and London.
However, with AI we need something even more innovative. I see methods like liquid cooling starting to become the ‘go-to’ choice. Beyond that, we’ll also start to see the emergence of phase-change cooling into the mainstream, a cooling method that harnesses a cooling fluid's natural latent heat of vaporisation, or the point at which it transitions from a liquid phase into a gaseous phase. Beyond that, you’re looking at building data centres on the top of mountains, but I think it’s safe to say we’re a little way away from that.
More data means more multi-cloud deployments
Data continues to grow in both value and volume. What used to be created in very centralised locations, such as cities, is now created everywhere. In our homes, walking down the street, you name it. This proliferation is making it harder for businesses to access, aggregate, and ultimately, make sense of their data. We call this the ‘Data Gravity’ effect, and it’s a trend that is only going to become more apparent as our dependency on data and digital services increases.
Businesses need customisability. They need to be able to access the apps and services needed to unlock value from their data all in one place, kind of like Apple’s app store. The data centre of the future is supporting this need with highly connected, geographically diverse platforms that enable businesses to reach the most strategic global markets with infrastructure that’s flexible and deployable in minutes. The importance of this approach is only going to become more apparent next year as enterprises recognise that multi-cloud deployments and the adoption of hybrid IT are the only ways to navigate a world driven by data.
If there’s one takeaway from this year, it’s that infrastructure is still rapidly evolving. Whether it’s finding new ways to power it through things like multi-factor power, upgrading it to support the technologies defining our future, or addressing the data-related challenges of tomorrow by overhauling it completely, fit-for-purpose digital infrastructure is needed now more than ever before.
Stay focused on the bytes and atoms.