The global AI economy is entering an electricity-intensive phase, forcing technology companies to secure power supplies alongside chips, computing capacity and data. For Nigeria, the shift creates both a warning and an opportunity: electricity must be treated not only as a social-development need, but also as economic infrastructure, digital infrastructure and a foundation of national competitiveness.
AI’s next constraint is access to power
For most of the past 10 years, the artificial-intelligence debate focused on algorithms, semiconductors, graphics-processing units and data. In 2026, electricity has moved to the centre of the discussion. Billions of dollars invested in GPUs have limited value if data centres cannot obtain sufficient and dependable power.
The world’s largest technology companies are consequently adopting strategies more commonly associated with energy businesses. They are signing nuclear-power contracts, funding new reactors, developing dedicated gas plants, acquiring or reserving power infrastructure, installing storage and choosing data-centre locations partly according to electricity availability.
The International Energy Agency projects that data centres worldwide will consume about 945 terawatt-hours of electricity by 2030, almost twice current usage. Demand is expected to rise by roughly 15% a year from 2024 to 2030, more than four times the rate for electricity consumption across the rest of the economy.
The agency describes this wider transformation as an “Age of Electricity”. It expects annual global demand additions between 2026 and 2030 to average about 50% above the yearly increase recorded during the previous decade. Data centres, industrial electrification, cooling, manufacturing and electric transport are among the main sources of growth.
For Nigeria, this trend should prompt a change in policy. The country needs to view power access as a platform for productive investment and digital expansion, rather than only as a social intervention.
How major technology companies are building their power strategies
1. xAI is developing power capacity around its computing facilities
The widely repeated claim that Elon Musk bought a nuclear power station appears to be wrong. There is no reliable evidence that Musk or xAI acquired an operating nuclear plant.
What the episode reveals about AI’s electricity needs is nevertheless significant. Musk said in 2025 that xAI had bought power-generation equipment abroad for use in the United States. The company has since moved aggressively to establish electricity supplies around its vast Colossus computing operation, rather than depending entirely on conventional grid connections.
Large numbers of gas turbines were deployed to support xAI’s Tennessee data-centre operations. Parts of that development have attracted environmental and permitting objections. In January 2026, xAI announced plans to invest more than $20 billion in its Southaven, Mississippi, data-centre project, which is expected to take the company’s computing capacity towards approximately 2 GW.
The key message is not whether Musk purchased a nuclear facility. It is that an AI company is effectively responding to slow or inadequate grid access by developing its own power infrastructure. That marks a major change in the relationship between the technology and energy industries.
2. Meta is becoming a major corporate buyer of nuclear power
Meta announced nuclear-energy agreements in January 2026 with Vistra, TerraPower and Oklo. The arrangements are intended to support as much as 6.6 GW of new and existing clean generation by 2035.
Including its wider commitments, Meta later said it was backing approximately 7.7 GW of nuclear capacity through agreements involving Vistra, TerraPower, Oklo and Constellation.
For a company whose core businesses are internet services and artificial intelligence, securing gigawatts of nuclear electricity has become part of its long-term infrastructure plan. AI companies are increasingly acting as anchor customers for new generation assets.
Nigeria could examine whether major digital businesses, telecoms operators, industrial parks, universities, mining projects and manufacturing clusters could serve a similar role for distributed power systems.
3. Microsoft is supporting the return of a retired reactor
Microsoft signed a 20-year power-purchase agreement with Constellation, helping to support the planned restart of Pennsylvania’s former Three Mile Island Unit 1. The facility is now called the Crane Clean Energy Center and is expected to bring about 835 MW of firm generation back into service.
The project also exposes the difference between having generation capacity and being able to deliver its output. Constellation has said the plant could begin producing electricity in 2027, while PJM has indicated that transmission work might postpone full connection to the grid until 2031.
In other words, a power station may be ready to operate even when the network cannot yet carry all of its electricity to customers.
The relevant performance question for future energy policy should therefore go beyond the number of megawatts constructed. It should ask how many dependable megawatts reach productive users, for how many hours, and what economic output those supplies generate.
4. Google is helping advance small nuclear reactors
Google reached a major agreement with Kairos Power covering advanced nuclear generation. In April 2026, Kairos began construction of its Hermes 2 demonstration facility in Oak Ridge, Tennessee.
The project is designed to provide as much as 50 MW to the Tennessee Valley Authority system and help reduce emissions associated with Google data centres in Tennessee and Alabama.
The broader importance lies beyond the facility’s 50 MW. By committing demand, Google is helping establish a future market for small modular and other advanced reactors, in much the same way technology companies helped expand the market for renewable power-purchase agreements in earlier years.
Large electricity users are beginning to influence which energy technologies receive financing and reach construction. Nigeria could use the same model for distributed renewable power: large buyers with dependable demand can provide the revenue certainty required to finance generation projects.
5. Amazon is tying data-centre expansion to nuclear capacity
Amazon has invested in X-energy, a leading developer of small modular reactors. Its initial agreement included support for manufacturing facilities capable of underpinning more than 5 GW of future nuclear projects using X-energy’s technology.
Investor interest has subsequently provided financial backing for the concept. In April 2026, X-energy, supported by Amazon, raised approximately $1.02 billion through a US initial public offering.
Amazon Web Services had earlier acquired a data-centre campus next to Pennsylvania’s Susquehanna nuclear facility for about $650 million. The transaction illustrates how computing assets are moving towards locations where electricity is already available.
- Available power increasingly influences where data centres are built.
- Data-centre locations affect where capital is committed.
- Investment locations shape employment and wider economic activity.
Electricity availability is therefore becoming a determining factor in the geographic distribution of digital capital.
6. Stargate illustrates the scale of the AI-energy merger
OpenAI’s Stargate initiative has set a target of approximately $500 billion in investment and 10 GW of AI infrastructure capacity.
In January 2026, OpenAI and SoftBank each committed $500 million to SB Energy. OpenAI also chose SB Energy to develop and operate a previously announced 1.2 GW data-centre site in Texas.
SB Energy’s role extends beyond constructing buildings. It combines data-centre development with the associated energy infrastructure. AI facilities and power assets are consequently beginning to be planned as a single, integrated class of infrastructure.
Demand, generation and the limits of the grid
7. Electricity consumption is accelerating
After years of relatively modest growth, US electricity demand is expected to set new records in 2026 and 2027. The Energy Information Administration identifies artificial intelligence and data-centre expansion among the principal drivers.
- 2025 consumption: approximately 4,195 billion kilowatt-hours.
- 2026 forecast: approximately 4,268 billion kilowatt-hours.
- 2027 forecast: approximately 4,391 billion kilowatt-hours.
The trend is global. The IEA estimates that electricity generation serving data centres could rise from roughly 460 TWh in 2024 to more than 1,000 TWh by 2030. The additional requirement alone would be comparable to the annual electricity use of some major industrial economies.
Power is consequently becoming a scarce strategic resource. Countries able to offer large volumes of dependable and competitively priced electricity will have a stronger advantage in attracting digital investment.
8. AI is bringing gas turbines back into focus
Although renewable-energy investment has reached unprecedented levels, AI systems require power around the clock. That need is helping drive a rapid expansion in natural-gas generation.
Siemens Energy reported record orders and financial results in 2026, partly reflecting demand for gas turbines linked to AI data centres and new power plants. Planned US gas-generation capacity rose sharply during 2025, while turbine producers faced exceptionally long delivery schedules. Some large machines are effectively reserved several years in advance.
The development highlights a practical distinction in the energy transition. Renewable sources can offer low-cost electricity, but high-value digital facilities also need availability, reliability, dispatchability and rapid construction.
For Nigeria, the appropriate response is not an ideological choice between individual technologies. The priority should be dependable, affordable power built from a combination of:
- Solar generation;
- Battery storage;
- Natural gas;
- Grid infrastructure; and
- Demand-management systems.
9. Network capacity may be the most serious bottleneck
Three days before the article’s reference point, PJM, the largest electricity market in the United States, proposed emergency measures that could require large data centres to switch to backup generation when the system nears critical operating conditions.
PJM serves about 67 million people. Yet even this highly developed power market is struggling to absorb the rapid growth of large new loads.
The emerging constraint is not always the absence of generation. In many cases, the limiting factors are transmission lines, transformers, substations, interconnection rights, grid flexibility, storage and permits.
Infrastructure once regarded as routine and unremarkable is becoming strategically important to the AI economy.
10. Renewable output is exceeding grid absorption in China
A development with particular relevance for Nigeria occurred on 17 August 2026. Estimates indicated that China curtailed about 360 TWh of renewable electricity during the first six months of 2026 because renewable generation expanded faster than sections of the grid could absorb it.
The same issue is becoming more visible in other markets. California curtailed approximately 4.5 million MWh of solar and wind power during the first half of 2026, already surpassing the total curtailed during all of 2025.
Storage is therefore becoming a more integral part of renewable-energy development. The central question of the future will not simply be how to produce electricity. It will also be how to store, transport, manage and commercialise power when generation is available.
The emerging formula for AI competitiveness
These developments point to a new economic equation:
- AI competitiveness = compute + electricity + land + connectivity + capital.
A country may possess strong AI talent but still struggle to participate fully in the technology economy if its electricity system is inadequate. Conversely, nations with abundant and dependable energy resources can convert power availability into a platform for attracting digital investment.
The competition is therefore no longer only about software engineers, chips or data. It is also about the ability to secure land, connect facilities, finance infrastructure and deliver electricity at the required scale and reliability.
Nigeria’s opportunity: building digital energy zones
Nigeria’s electricity shortfall is currently a major development constraint. It could also become an investment opportunity because much of the country’s power infrastructure remains to be built.
That starting position may allow Nigeria to bypass some legacy approaches. Instead of concentrating exclusively on a sequence of central generation, long-distance transmission, distribution networks and final consumers, the country could plan energy and productive demand together in defined locations.
A more integrated model would combine the following elements within the same geography:
- Power generation;
- Energy storage;
- Digital infrastructure; and
- Productive electricity demand.
Such locations could serve data centres, industrial users, telecoms infrastructure, universities, mining operations and manufacturing clusters. Guaranteed or large-scale buyers would help create the demand certainty needed to finance new generation, while co-locating power and users could reduce transmission bottlenecks.
Moving from a focus on installed megawatts to genuine economic sovereignty requires a fundamental rethink of national strategy. For the Rural Electrification Agency, RAMCO and other institutions, the next step is to develop a blueprint for Nigerian Digital Energy Zones in which power supply, storage, connectivity, investment and productive activity are planned as one system.
