Artificial Intelligence is driving up demand for electricity, so how can you make the most of the financial opportunity, asks Luke Staden, founder of Staden Financial Management.

In late 2021, after China banned crypto mining, Kazakhstan became the world’s second-largest mining hub, drawn by cheap, subsidized electricity. This sudden, massive, and largely unregulated influx of miners overloaded the nation’s aging power grid, consuming up to 8% of its total generating capacity.

This energy drain became a tipping point. When mass protests erupted in January 2022 over unrelated fuel price hikes and corruption, the overloaded grid failed, plunging communities into the freezing winter dark. The power shortages fuelled public rage, linking the miners’ greed to the public’s suffering. Culminating in the deadliest period of civil unrest in the country’s post-Soviet history.

Kazakhstan’s problem was that crypto miners converted energy into profits. As long as energy costs were lower than the value of crypto generated, it made economic sense to use as much energy as possible.

Data centres

I’m reminded of Kazakhstan because as we see data centres start to be built to support the growth of AI in the world, it seems that we once again have assets which convert energy into profits.

AI queries cost 10 times the data than a standard internet search engine. AI is incredibly power intensive. As we see more and more AI infrastructure created we will see more and more energy demand.

This massive, instantaneous demand requires utilities to spend billions to upgrade power grids. In the US utilities are regulated monopolies, they are legally permitted to pass these costs (plus a guaranteed profit) directly to ratepayers.

Energy costs

The rising electricity bill is the financing vehicle for the next wave of energy investment. While the European system is more nuanced you can be sure that such an increase in energy demands will also lead to an increase in energy bills.

The bottom line is that AI is likely to increase energy costs for everyone in a significant way. I don’t think there’s any way to prevent CEOs across the world trying to replace their work force with AI so it seems fairly inevitable.

My poorly thought-out method to cope with the existential threat of AI replacing swathes of employees across the world is to work out ways to try to profit from it happening.

Investment options

Personally I see four investment options that can take advantage of the increased AI energy consumption.

The most secure option is to invest in US utility companies. When a data centre is built, US utility companies are legally obligated to provide service, forcing them to undertake massive capital expenditure for new power lines and substations.

However, regulators guarantee the company a steady rate of return on spending. The more they spend the more revenues they are guaranteed. The faster AI forces them to invest, the faster their asset base grows, ensuring predictable, non-volatile earnings growth.

The highest growth investments however are likely to be into investments which play a role in power generation. The three options to consider are natural gas, renewable energy and nuclear power.

Other energy sources

Natural gas is the essential short-term bridge. Data centres need power in two years, but new solar and nuclear plants take longer to permit and build. Gas-fired generation can be brought online fastest, making it the unavoidable “starter fuel” for many new AI hubs.

Renewable Energy AI companies are publicly committed to clean energy. This means they are signing massive, multi-decade power purchase agreements to back their new data centres.

I’m looking at  utility-scale renewable energy developers and, critically, companies that produce grid-scale battery storage. Unfortunately, renewable energy is currently held back by its inability to have on demand energy. Solar doesn’t work at night, Wind doesn’t work with no wind etc.

Nuclear power

Nuclear is the long-term solution for reliability. AI processing cannot tolerate even a flicker of interruption. Since nuclear power delivers continuous, carbon-free baseload energy, it is the best fit for this critical need. We are seeing major tech companies directly investing in Small Modular Reactors (SMRs), which offer a high-upside opportunity to power future data centres directly.

In my typical “cover all bases” approach, rather than pick one strategy to capitalise on increasing energy prices, I will be using them all. Well I’m still in two minds about natural gas as it’s probably going to perform the best short term but long term it has the cost of not being able to tell my children I had no part in destroying the environment.

Contact Luke Staden of Staden Financial Management to set up a meeting and discuss your particular financial needs or check out the YouTube channel.