According to the United Nations Environment Programme (UNEP), a single large language model (LLM) query — from the likes of ChatGPT, Gemini and Claude — requires 2.9 watt-hours of electricity, compared with 0.3 watt-hours for a regular internet search. Training a single LLM results in approximately 300,000 kilograms of carbon dioxide emissions, about the same as five times the lifetime emissions of an average car or comparable to 125 round-trip flights between New York and Beijing. While these estimates have their limitations, one thing is for sure — AI is going to need a lot of electricity.

You’re probably looking at the tab with your favourite AI chatbot open on it with slight horror right now, aren’t you? Who would’ve suspected that it consumes so much energy? Not to mention the gallons upon gallons of water needed for the production of these technologies.

Electricity usage by data centres worldwide is expected to double within the next two years, potentially matching Japan's current level of electricity consumption. The good news is that electricity can come from renewable energy sources — in fact, by 2050, about 90 per cent of it.

Some places are already making it happen. In its Northern European-based data centres, Google claims that more than 90 per cent of the energy came from renewable sources. The percentage falls quite drastically in Asia and the Pacific, however;  less than 20 per cent is renewable energy. It is no coincidence that Northern Europe has one of the most interconnected power markets in the world, through Nord Pool. Nord Pool – which covers the Nordic and Baltic regions – is an electricity market where energy is bought and sold. It allows electricity to flow more freely across borders, ensuring that regions with excess power can supply regions with shortages, and helps to balance supply and demand in a more cost-effective way.

“If in one country, there’s more wind, the prices drop and then it allows this country to export electricity at lower prices to neighbor countries within one day, in real time,” says Anna Lobanova from the Energy Division of the United Nations Economic and Social Commission for Asia and the Pacific (ESCAP).

Connecting power systems across borders, both within and between countries, opens up greater access to renewable energy resources and helps integrate them more effectively into the overall energy grid. Power systems with more renewable energy face greater fluctuations in generation compared to those using fossil fuels. Connectivity helps manage this by linking larger regions. For example, solar power shifts throughout the day, so connecting different areas can extend its availability. Wind patterns also vary, and larger, connected systems can use wind power wherever its strongest, smoothing out the fluctuations.

In Central Asia, ambition has been increasing in restructuring the energy sector. Traditionally reliant on fossil fuels, Kazakhstan is aiming to have half of its energy mix come from renewables by 2050.  Meanwhile, Uzbekistan set a goal to have a quarter of its energy come from renewables by 2030.

But, as Iva Brkic from the United Nations Economic Commission for Europe (UNECE) points out: “We will need a more resilient and robust power system to absorb this additional renewable energy capacity.”

ESCAP, jointly with UNECE, has been working on promoting an integrated power system in Central Asia, leveraging the connections between the region’s Soviet-era infrastructure. The key being deepened regional cooperation, allowing countries to formulate more ambitious emission reduction strategies. But this goes beyond just Central Asia. If we hope to see more energy come from renewable sources, especially as the demand for electricity rises, power connectivity may be the solution — in the Asia-Pacific region and beyond.

So, yes, if you want to keep asking your AI chatbot for help with whatever task you need, we need a more sustainable, reliable and cost-effective energy system. For that, we need to work together.