AI-fuelled power demand and the tech revolution will give new life to natural gas [Gas in Transition]
The explosive growth of AI and tech revolution is leading to an equally rapid expansion in the setting up of data centres to handle and process the vast amount of generated data. Data centres are energy-hungry and their power needs are set to soar.
Microsoft alone is opening a new data centre globally every three days.
Data centres need reliable, uninterruptible power to function 24/7. Renewables, even when backed-up by batteries, cannot provide that on their own. While this remains a problem, natural gas is the only alternative fuel offering reliable, steady, power at scale, with almost 50% less emissions than coal.
The International Energy Agency’s (IEA) estimates that power demand from data centres globally will double by 2026, in comparison to 2022 levels, to over 1000 TWh, and carry on increasing rapidly. This is equivalent to about a quarter of US total power consumption in 2023.
Data centres
The need for and the continued growth of data centres is spreading globally, in response to user needs. It is estimated that the capacity needed to serve AI-focused deployments will triple by 2030. New technologies like generative AI –considered to be a technological innovation comparable to the Internet – machine learning, streaming, cloud gaming, blockchain and virtual reality generate vast amounts of data daily, usually stored in remote servers in data centres constructed all over the world.
Every online interaction relies on information stored in such remote servers. “The data centre industry is the backbone of AI and just about everything else digital,” from search engines to e-commerce.
According to a recent report, NVIDIA alone is expected to be shipping 1.5mn AI server units per year by 2027. These, running at full capacity, could consume over 85 TWh/yr, almost three times the annual energy consumption of Ireland. NVIDIA estimates that “the total value of all the gear in data centres will rise to $2tn in the next four or five years.”
Another staggering statistic is that, according to the IEA, since 2010 the number of internet users worldwide has more than doubled, while global internet traffic has expanded 25-fold. In 2022 global internet traffic was 30% up on 2021.
Data centers are “typically owned and operated either by big companies, such as cloud vendors, banks, telcos, but also by ‘co-location companies’ that lease out space. New companies are also emerging to service these needs, such as American EQT Infrastructure that has partnered with EdgeConneX to develop data centres for customers globally to “support their future cloud, AI and other critical digital infrastructure requirements.” They see this as a market with massive growth opportunities. As of now, they have “a global footprint of 80 data centers in operation or development in more than 50 markets across the Americas and Europe.
According to McKinsey, the US data centre industry is expected to grow 10% per year through 2030. Demand for data centre capacity is so high that it has been outstripping supply since 2022.
Energy needs
The rapid growth in new technology and data centres requires vast amounts of energy. According to the IEA, worldwide, data centres and data transmission networks consumed between 2% to 2.8% of global electricity use in 2022 (excluding crypto) and this demand is growing at a very fast rate. In the US, 2,700 data centers consumed more than 4% of the country’s total electricity use in 2022. This is projected to exceed 6% by 2026.
This is after improvements in energy efficiency helped moderate growth in energy demand from data centres and data transmission networks. But despite strong gains in efficiency, according to the IEA “the rapid growth in workloads handled by large data centres has resulted in a substantial increase in energy use over the past several years, growing by 20% to 40% annually. Combined electricity use by Amazon, Microsoft, Google, and Meta more than doubled between 2017 and 2021, rising to around 72 TWh.”
AI is transforming the global economy. Technology innovations are coming thick and fast. New technologies and AI applications and their use are proliferating, putting even more pressure on availability of reliable and dependable energy. For example, “AI graphic processing units consume four times as much power as servers used for cloud applications.” New data centres “require exponentially more power than traditional data centers” – it is a paradigm shift. Generating this additional power will require new infrastructure, power plants, transmission lines, etc., as well as dependable energy sources. And as “AI becomes fully integrated into all facets of society,” such as manufacturing, transportation, and healthcare, “the energy requirements will skyrocket.”
In the US, AI and the growth in clean-tech manufacturing are already pushing power grids to their limits, with utilities struggling to keep up. Some parts of the US project electricity demand to increase by more than tenfold by the end of the decade. There are estimates that by 2030 this could represent 20% to 25% of total US electricity demand. The speed at which this is happening is already making sourcing and accessing power supplies the critical factors in new developments. Elon Musk said in April that the latest bottleneck to the cutting-edge technology was “electricity supply”, echoing Amazon warnings that there is “not enough energy right now” to run new generative AI services.
And it is all happening at a time the world is shifting “to low-carbon energy generation that would inevitably necessitate a decline rather than an increase in energy demand. This is exactly why the oil and gas industry is counting on the AI revolution to give new life to natural gas. Nuclear could be part of the solution but it takes years to build, so it is not “a viable option when AI companies need power immediately.”
This is what prompted Bloomberg’s Liam Denning to say “the AI tevolution needs chips, software — and gas pipelines.” Many AI executives now say that clean energy may not be able to meet their needs and are turning to fossil fuels.
With commitments to reduce emissions, ensuring 24/7 power supply to data centres – and adequate redundancy – will require a combination of renewable power and constant sources of energy such as natural gas. In many cases this will lead to delaying the retirement of fossil fuel plants and bringing new ones online.
Unless infrastructure and energy availability are expanded, these will become serious bottlenecks with time. Grid limitations are a challenge in Europe too.
Challenges
AI technologies and applications process vast amounts of information. "The more information they gather, the smarter they become, but the more information they gather to get smarter, the more power it takes." This is happening so fast that it poses major challenges to an energy industry used to a slower pace of development.
Sam Altman, CEO of OpenAI, summed up the challenge at Davos. He said, “I think we still don’t appreciate the energy needs of this technology.” He does not see how these needs could be met “without a breakthrough.” He went on to say “we need fusion or we need, like, radically cheaper solar plus storage, or something, at massive scale—like, a scale that no one is really planning for.” But that will take time and until that happens natural gas is the solution.
With economic growth increasingly and closely interlinked with growth in AI and new technologies, any limitations that inhibit this will also hamper economies globally.
However, the future of a massive number of data centres powered by natural gas is not without problems. Many companies, including Google and Microsoft, are committed to net zero by 2050 and use of clean electricity to power their future operations. This could pose a threat to the view that AI and the tech revolution will give new life to natural gas.
At some point in the future, reliable clean power will become available and it will replace natural gas. But, at least in the foreseeable future, that eventuality appears to be going further and further away in time.
Natural gas has a bright future well into the next decade. This is supported by most credible forecasts, apart, perhaps, from the IEA.