Artificial intelligence, climate and temperature

Note: With the release of new scenarios from Shell, please read here what scenarios are, and what they are not.

In early February the European Copernicus climate service announced that the January 2025 global average surface temperature was the highest ever recorded for that month. This came despite a shift away from the warmer El Niño conditions in the Pacific that are often associated with record breaking months, and which had been an underlying contributor to the yearly temperature anomaly of 1.55°C recorded for 2024. These recent temperature measurements raise the question as to where and when the global average surface temperature might settle in the decades ahead, given the ongoing rise in anthropogenic fossil fuel CO₂ emissions, as seen, for example, from 2023 to 2024. A casual inspection of recent emissions data gives a first indication of no change in direction, but beneath the headline trend of rising fossil fuel emissions, a very different story is beginning to emerge. To see that trend and imagine the changes ahead requires considerable analysis and a view about future societal trends.

So, we come to The 2025 Energy Security Scenarios, released by the Shell scenarios team last week. Three scenarios are offered, two which explore the future based on underlying trends playing out across the world today and one which takes a normative approach to understand the full depth and speed of change required to achieve net-zero emissions in 2050 and to limit warming to 1.5°C by 2100. All three scenarios recognise that the world is currently being shaped by security, competitiveness, and climate concerns, with technologies like artificial intelligence rushing headlong towards us.

The two exploratory scenarios are called Surge and Archipelagos.  In Surge, an era of robust economic growth is ushered in by artificial intelligence technologies (AI) that are welcomed and not overly challenged, with economic growth and AI infrastructure driving up energy demand. The geopolitical landscape offers a spur for change as China and the USA compete for AI dominance. The Archipelagos scenario sees a world still mindful of the energy system disruption in 2022 followings Russia’s invasion of Ukraine, but also a world that reacts to the pressures of increasing migration across multiple borders and uneven global trade patterns. Trade friction and geopolitics impinge on the speed of the energy transition, but this is countered somewhat by growing pressure to address climate change, which forces action across society, but not at the pace needed for achievement of net-zero CO₂ emissions in the nearer term.

Horizon, our third scenario, is normative and illustrative of a rapid acceleration of the energy transition and introduction of carbon management practices to sharply reduce emissions, both in response to a comprehensive policy framework with strong societal and political support. Horizon includes a fast and comprehensive change in global land management practices, including an end to deforestation in the 2030s, in combination with strong government support for the full range of carbon capture and storage technologies.

The 2025 Energy Security Scenarios are built on the back of The Energy Security Scenarios released in 2023, but by including Surge the scenarios team has explored the implications for energy in a world where AI begins to reshape society. This has a real impact on the energy system for several reasons. Most importantly, it sees energy demand increase due to greater economic growth and it transforms the way energy infrastructure is built in the decades ahead.

For two centuries the energy system has largely evolved through the development of big, bespoke projects, like offshore platforms extracting oil or the many huge refining complexes that dot the globe. But a trend that has emerged in recent years is to build equipment on assembly lines and assemble it, ‘Lego-like’, in the field. Solar PV is like this, and grid electricity storage has scaled up rapidly on the back of battery production. In Surge, this trend spreads and accelerates and the general technology push enabled by an AI world supports this, both at the manufacturing facilities and by enabling virtual networks of otherwise disconnected devices in the field.

Shown above, assembly line production of small modular direct air capture units in the late 2030s illustrates the change in the way energy infrastructure is built and deployed. Modular production accelerates the transition and AI systems ensure efficient integration of multiple units in the field.

The trend towards modular production benefits solar PV, electricity storage, heat pump deployment, hydrogen production via electrolysis, direct air capture of CO₂ and the introduction of small modular nuclear reactors (SMR). Other energy system technologies also benefit. In Surge, the introduction of SMRs via this route eventually transforms the marine sector and takes it away from liquid and gaseous fuels like marine diesel and LNG. The first SMR ships come into service just before 2050 and by late in the century in Surge, all ocean-going vessels are nuclear. However, this development isn’t seen in Archipelagos or Horizon.

Surge also sees a new business model for carbon management. A trend that has emerged in recent years has resulted in considerable venture capital funding for the development of direct air capture (DAC) technologies, with companies such as Google, Airbus, Microsoft, BCG and NYK all agreeing to buy future DAC carbon credits. Dozens of DAC startups have appeared and projects, albeit modest in scale, are underway. In Surge this trend accelerates as the cost of DAC drops rapidly with assembly line production of capture modules, akin to building air conditioners or refrigerators. The voluntary carbon market flourishes, with DAC credits underpinning it, and high credibility carbon neutral labelling enters widespread use. Both consumers and business customers demand carbon neutral goods and services. One outcome is that by the late 2040s global DAC with geological storage (DACCS) use exceeds traditional CCS within power generation and industry. The future of geological storage of CO₂ becomes a DAC story, with billions of tonnes per year of CO₂ captured and stored via this route later in the century.

Surge offers insight into a higher growth pathway, something which many governments are striving for. The increased growth means greater energy demand, but as discussed it also brings with it a faster transition and rapid scaling of carbon storage. Surge therefore reaches net-zero CO₂ emissions well within this century, meaning that by 2100 surface temperature warming is limited to 2°C, albeit after a temporary overshoot of this threshold.

But even in a slower growth, more fractious world, illustrated by Archipelagos, the momentum in the energy transition is sufficient to reach net-zero emissions by about 2120, which delivers about 2.2° of warming. The implication of this, in combination with Surge, is that warming is unlikely to exceed 2-2.5°C, a significant shift from multiple reports in the early 2010s where scenarios that delivered 4°C of warming were highlighted in adaptation analyses. Evidently, in just a decade, the current momentum in the energy transition has delivered a new global warming paradigm for consideration.

The energy transition is now at a tipping point; the questions at hand are not whether society can reach net-zero CO₂ emissions or whether society is able to limit surface temperature warming, but how soon will net-zero emissions be realised and just how low might the eventual temperature plateau be.