Cloud computing and data storage providers’ commitments to transition to net-zero operating models could make renewable and low emission energy technologies become more like traditional infrastructure assets.
In March of 2024, Amazon Web Services (AWS) struck a deal to buy a 960MW data centre campus in Pennsylvania, the United States, from Talen Energy.
What was remarkable about the deal was that the data centre was located right next to a nuclear power station, owned by Talen, which provided the centre directly with electricity. Under the terms of the deal, AWS has now committed to a 10-year power purchase agreement and is planning to double the capacity of the data centre.
The transaction could be a sign of things to come, as so-called hyperscalers – large cloud computing and data management companies, including AWS, Google Cloud, Microsoft Azure and Alibaba Cloud – try to secure sources of emissions free energy to facilitate their energy transition to a net zero economy and power further growth.
And as these hyperscalers become more dominant voices in the net zero energy sector, they could help new and early commercial low emission energy technologies become more like traditional infrastructure assets with their associated stable income streams.
“With the hyperscalers in particular, but actually the whole data centre sector, we have a group of people who don’t want to take years to do these projects. They want to take months. And they’re putting in place some of the fuel policies that enable that to occur,” Kate Vidgen, Global Head of Industrial Transition and Clean Fuels, Macquarie Asset Management said at the JANA Annual Conference in Melbourne in early September.
“They’re happy to do multiple product deals across multiple jurisdictions, so you get your scale, and they’re happy to support nascent industries. So those industries that we find really hard to get off the ground, they’re actually helping.
“What this means is they’re actually de-risking projects for institutional capital, making it more infrastructure-like and making it easier for debt and equity to invest,” Vidgen said.
Hyperscalers are among the companies with the most ambitious emission targets. For example, Google has set itself a goal to achieve net-zero emissions across all of its operations and its value chain by 2030.
The search behemoth aims to reduce 50 per cent of its combined Scope 1, 2 and 3 absolute emissions by 2030, compared to its 2019 base year.
Vidgen believes companies like Google will come to play a key role in the renewable and low emission energy sector.
“The hyperscalers are the first people in line because they’ve had a track record of signing large contracts, and the reason they’ve done that is they’ve set really ambitious goals. They want to do a lot of this, most importantly, with 24-7 [access to] power,” she says.
“Microsoft has gone to the next step. Microsoft wants to be net negative emissions by 2030. And by 2050, they want to offset all the historical emissions. So that’s a pretty big target,”
Vidgen said.
The bigger and more sophisticated data centres are starting to think about moving their services to locations where they may have access to a greater supply of energy, at different points in the day.
This could ultimately become a multi-continent approach, Vidgen speculated, especially as these companies are looking for access to energy around the clock.
“I think the dream, which potentially could become a possibility, is you actually follow the wind and the sun. You skip continents by providing services where you’ve got energy. So over time, we may be able to do that,” she said.
Blame A.I.
There has been some concern that the current advances in artificial intelligence (A.I.) are set to cause a big spike in the demand for electricity, as companies require a lot of energy to train and operate their new A.I. models.
And although there is a clear increase in energy demand noticeable in recent years, Vidgen is somewhat sceptical that this is driven by A.I. alone. For example, the decision of many companies to reshore their manufacturing capabilities following the COVID-19 pandemic has caused an uptick in energy usage in developed markets.
The global energy transition more broadly, including the electrification of large parts of the economy, such as transport and heating, necessarily means there will be further growth in the demand for electricity.
But the impact of A.I on energy demand is far from obvious, Vidgen said.
“The last decade, the data centre’s electricity demand growth was pretty flat. Yes, there were things like the advent of cloud computing, but the hyperscalers came along and they tend to be more efficient because they’re scaled and they can create efficiencies through that. We saw equipment being more efficient within data centres itself,” she says.
“We think that we will have some demand [from A.I.] come through very quickly, but we don’t really know what that means in the long term.
“What we do know is that chips are likely to become more efficient. We know that the way we train our models, which is pretty energy intensive, will change and adapt, and actually focus on efficiency.
“We’ve seen it before: when there is a resource constraint, industries change their business models, they change their technology, and they change their services,” she said.
Vigden points to the creation of the internet and even the invention of the personal computer. When these came along, people also warned of a widespread explosion of electricity usage, but this never eventuated.
“Forbes came out in 1999 and had a headline which was: ‘Dig more coal, the PCs are coming’. In fact, some people were saying they thought within 10 years, half the US grid would be required to power the internet,” she said.
“That didn’t happen. We didn’t see a massive explosion [in energy demand]. What we actually saw was pretty moderate growth in electricity demand for the first decade of the century,” she said.
Besides, the fact that the electrification of the economy will lead to a higher demand for electricity is hardly a revelation, Vidgen argued.
“It’s a bit funny that everyone’s getting really excited about increased electricity demand. We knew this was coming; it’s not only A.I. The energy transition will require things like electric vehicles. It will require heat pumps and we have to electrify industry,” she said.
“The International Energy Agency has been saying for a long time that electricity demand is probably going to double in their most conservative case, so we knew this was coming. That means it shouldn’t be a surprise to us that we need to grow our own infrastructure.”
__________
[i3] Insights is the official educational bulletin of the Investment Innovation Institute [i3]. It covers major trends and innovations in institutional investing, providing independent and thought-provoking content about pension funds, insurance companies and sovereign wealth funds across the globe.