After decades of decelerating growth in electricity demand, we see a significant shift on the horizon. The increasing electrical requirements for the electrification of mobility – such as cars, trucks and other vehicles – as well as homes and industrial facilities, are likely to drive an acceleration in electricity demand. Additionally, the substantial power needs of generative AI will further steepen the demand curve. A key focus for us has been estimating the amount of electricity we may need by 2030 and ensuring there will be sufficient supply to power this new electrical revolution.

What does the demand picture look like?

Access to reliable electricity is something we take for granted. The power grid and its constellation of generating stations were largely built in the mid-20^th century in North America. This massive infrastructure project has since enabled broader use of air conditioning, ubiquitous computing, the growth of the internet and many other conveniences that define modern life. 

After the initial build-out came a lengthy period focused on efficiency. Engineers went to work on every area of demand, cutting electricity consumption of individual loads on the system significantly over the past 40 years. This includes more efficient lighting, home appliances and computers. As efficiency increased, less investment was required for new generation capacity or grid upgrades. Despite significant economic growth over the last two decades, U.S. electricity consumption has increased by only 0.5% per year, on average, over the last 20 years.¹

There is increasing evidence that electricity demand growth has started to accelerate. New demand vectors – many with durable secular growth characteristics – are driving this inflection. One significant factor is the impact of AI.

As outlined in an article by the U.S. Equity Growth team published last year, we believe we are in the early stages of a major technological platform shift, with rapid growth in capital expenditures to support generative AI. Much of this growth will go toward building new data centers. We already see this in capex and leasing data, which implies a significant acceleration in planned data center construction.

The data centers needed to power generative AI workloads will likely drive new electricity demand growth higher, given their reliance on more sophisticated processors, such as GPUs (graphics processing units), which can be far more power hungry. For example, the IEA estimates that AI-based searches use nearly 10 times the electrical power of standard Google searches (2.9 watt-hours compared with 0.3 watt-hours).² In addition to the significant direct power consumption increase, more sophisticated cooling solutions may also be necessary given the more intense heat emission from GPU server racks, further driving power needs in data centers.

Overall, data centers could consume more than 9% of total U.S. electricity by 2030, up from just 4% today.³ This expected increase in power usage from data centers adds to the already growing power demand related to electric vehicles and broad electrification trends, such as heat pumps.

Will supply be there to meet demand?

Will the availability of electricity be a gating factor around the growth in electric mobility, electrified homes and industrial facilities? Will it impact the development and use of AI? We can make a case for power supply sufficiency through the end of the decade, but we have concerns around electricity supply sufficiency in the 2030s.

In the U.S., there has been a modest level of investment in overall electrical capacity over the past 10 years, but this appears to be changing. Seeing the coming demand wave, utilities and other power providers are adjusting their capital spending programs upwards to accommodate expected future load growth. If utility companies continue to make these investments, then we do not believe that power will be a gating factor for AI growth. The 2030s pose a more challenging question, however. If AI applications become ubiquitous in our daily lives, electrical consumption from data centers could require greater investments in power generation capacity.

Utilities and independent power producers will likely need to build significant amounts of new power generation capacity with an “all of the above” approach. This includes renewables and storage as well as new natural gas. There will also be a pressing need to extend the life of existing generating capacity and increase electrical grid capacity. Additionally, we expect that nuclear power generation will also see a resurgence in the 2030s. However, we believe expecting new nuclear power capacity to arrive sooner than that is overly optimistic given the technological, cost and siting challenges. Finally, we are also watching a host of innovative power generation and storage technologies that could help us bridge the gap between demand and supply in the 2030s – such as hydrogen, geothermal, wave energy, fusion power, advanced solar tech, thermal and solid-state batteries and next-generation conductors for the grid.

Where are we looking for opportunities to capitalize on the electrification trend?

We see opportunities in four key areas:

1. Renewables and storage: Attractive government subsidies in countries around the world continue to push the cost of renewables lower. As a result, we expect to see higher growth and market share shift toward renewables from traditional energy sources, even as all sources of energy appear likely to grow.
2. Traditional power generation: We believe in an “all of the above” approach to energy supply and transition over time. This means continuing investments in new natural gas generating capacity. We will also need to commit to maintaining and extending the life of existing power stations, which may need to run longer than initially expected. We are also exploring potential applications of natural gas power generation “behind the meter.” For example, a data center operator striking a deal to develop and operate dedicated power generation capacity for its data centers. This unconventional approach could circumvent long waits for power connections, particularly in key data center markets.
3. Grid expansion and renewal: Grid investment is the enabler for both points above. If we cannot connect our new renewable energy capacity to the grid in a way that improves resilience in the electrical system, grid stability could decline. Grid expansion is also the cheapest and most effective way to mitigate renewables’ intermittency disadvantages. This will require investments in upgrading the existing grid as well as building greater interconnectivity between regions and power-generating assets.
4. Electrical infrastructure upgrades: In a more electrified world, we expect companies that make technologies and components enabling homes, businesses and data centers to consume greater electrical loads will be long-term winners.

Investment implications

We expect a wide range of investment opportunities across companies and industries including utilities, capital equipment, construction and materials that may benefit from this acceleration in electricity demand in various ways. The biggest winners will be those with the assets and managerial quality to navigate this growth profitably, while others may struggle with the inflecting demand. An active approach that can adapt as those winners and losers emerge and evolve over time will be crucial for capitalizing on this electrification wave. 

^1 ”Total Energy Monthly Review,” U.S. Energy Information Administration; data as of June 2024.

² “Electricity 2024 - Analysis and forecast to 2026,” International Energy Agency (IEA), iea.org; data as of January 2024.

³ “Powering Intelligence: Analyzing Artificial Intelligence and Data Center Energy Consumption,” Electric Power Research Institute (EPRI), epri.com; data as of May 2024.