A significant boost to future global electricity demand from the construction of power-hungry data centres processing artificial intelligence (AI) applications is now a well-established expectation. But, as BP points out in its latest annual energy report, the impact of AI on the energy sector goes a lot further than that.
In its Energy Outlook 2025, the supermajor says that demand from data centres will provide a significant boost to electricity demand, but the extent of that impact will vary from region to region. So, while the increase in data centres’ electricity usage is projected to comprise around 10% of overall global power demand growth up to 2035, in the US, data centre demand could account for as much as 40% of national power demand growth over same period.
BP notes that these forecasts are highly uncertain, because they depend on how fast AI is adopted – and on how energy efficient data centres can become. The report notes that global digital data traffic increased by 25 times in the 2010-14 period, while the energy use of data centres only rose two-fold in the same time.
However, the factors that have kept energy demand in check may not be replicable in future – for example, the process of moving data that used to be stored in data centres to more energy-efficient cloud storage has largely been completed. However, BP says that “other innovations, including continuing advances in chip design and AI programming, are likely to drive more improvements in the energy efficiency of data centres in the coming years”.
Just as interesting as the impact of AI on power demand at data centres are other potential impacts of AI on the energy sector, particularly on the supply side.
AI is helping to make building and operating energy systems and power grids more efficient, for example, through more accurate forecasting of demand spikes that facilitate switching between power sources of all types from rooftop solar to gas-fired power stations.
As attendees at recent Future Digital Twin & AI conferences can testify, the pace at which AI is being adopted in the hydrocarbons industry is also speeding up, enabling faster analysis of field geology through more accurate prediction of equipment failure to improving the efficiency of supply chains. AI is also benefitting clean energy generation, for example by helping in the development of new materials for solar panels, better batteries and more efficient hydrogen production.
“More radically, continued advances in AI could unlock major technological breakthroughs in low carbon energy supply, such as through the development of new advanced biofuels, or even of usable nuclear fusion,’ BP says.
Forecasts are further complicated by uncertainties over the impact of AI in the wider global economy: how far will it enable the automation of tasks, or enable better use of physical resources, or stimulate innovation? AI is likely to improve industrial efficiency, make transport more fuel efficient through better traffic management and route optimisation, and make heating and cooling of buildings more efficient. But by how much? Such variables make it hard to come to firm conclusions on the impact of AI on energy demand.
Contrasting scenarios
Another major uncertainty for the energy industry is the extent to which the world is prepared to tackle global warming. To tackle that unknown, the outlook provides forecasts based on two scenarios. The Current Trajectory scenario extrapolates along the “broad pathway along which the global energy system is currently travelling”, while the Below 2° scenario looks at how different elements of the energy system could achieve substantial carbon emissions reductions.
However, in either scenario, the short-term picture remains broadly the same. One standout set of data is that BP is now forecasting that peak global oil demand will not happen until around 2030, reaching some 103m barrels a day by then – and it could even be a few years later than that, if energy efficiency gains fail to kick in as expected.
Just a year ago, the firm was predicting peak oil would happen as soon as 2025, but it says slowing energy efficiency gains are among factors that have helped push the date back.
After 2030, the current trajectory forecast shows a relatively slow decline in oil production to 83.8m b/d by 2050. In the Below 2° scenario, output in 2050 is forecast to fall to 33.8m b/d.
This data provides another indication of just how volatile the mechanics of the global energy system are at present and how difficult it is to make accurate predictions about its future.
(Photo: Shutterstock)
