Artificial intelligence is turning electricity into the new industrial currency

This article is part of In conversation about sustainable finance & emission reduction systems, a new series by Diego Balverde. You’re reading volume 25 of the Breaking news series. Here is volume 24

Artificial intelligence is ceasing to be only a technological revolution and is becoming a physical reorganisation of the global economy because every new model, every agent, every data centre, every industrial automation system and every computing platform requires energy infrastructure that cannot expand at the same speed as software, and that difference between digital speed and electrical speed is creating one of the largest investment, employment, infrastructure and financing markets of the coming decades.

The real bottleneck for artificial intelligence is no longer located only in chips. It is located in available electricity, transmission lines capable of moving it, transformers capable of delivering it, substations capable of distributing it, storage capable of shifting it, cooling systems capable of sustaining it, permits required to build it and capital willing to finance all of that infrastructure before digital demand collides with the physical limits of the power system.

Developments across energy markets are making increasingly clear that the AI debate is moving from semiconductors toward generation, grids, storage and firm power because large data centres need to operate continuously and cannot depend exclusively on a single intermittent source, meaning that nuclear, flexible gas, hydropower, renewables, BESS, transmission and intelligent demand management are beginning to form one strategic portfolio.

The mistake would be to turn this transformation into another ideological argument over which technology must win because no isolated technology can answer demand that combines rapid growth, continuity requirements, grid congestion, environmental constraints, industrial competition and pressure on the cost of capital.

Nuclear power is regaining importance because it offers firm low-carbon generation over long operating periods, but building new plants requires time, permits, supply chains, fuel, technical knowledge and capital. Gas continues to provide flexibility because it can deliver dispatchable capacity with shorter development timelines, although exposure to international prices and emissions means it should operate inside a broader architecture rather than serve as the only solution.

Solar and wind continue offering highly competitive marginal generation and can add enormous volumes of electricity, but they require grids, storage, interconnection and flexibility to transform variable production into power usable by facilities operating twenty-four hours a day. Hydropower retains exceptional value where available because it combines generation, flexibility and in some systems storage. BESS can shift electricity, respond within seconds, reduce peaks and protect constrained nodes, but cannot independently substitute long-duration firm generation.

The real opportunity lies in integrating all these assets as a system and understanding that artificial intelligence is creating not merely electricity demand but an entirely new economy around electricity. A region capable of offering hundreds or thousands of megawatts of firm capacity, a robust grid, storage, available land, reasonably predictable permitting, water or alternative cooling solutions, fibre connectivity and access to capital can become an artificial-intelligence hub even if only a few years earlier it would never have been considered a technology capital.

The opposite is equally true. A city with universities, talent, technology companies and connectivity can lose multibillion-dollar investment if it cannot guarantee electricity. Territorial competitiveness therefore begins to be measured in available megawatts, connection times and infrastructure quality. Electricity becomes the new industrial currency because it determines which economic activities can locate in a region, how quickly they can grow, how much they cost to operate and what type of capital that region can attract.

This has profound implications for employment because the common narrative that artificial intelligence will simply destroy work observes only one part of the system. Automation will eliminate or reduce certain repetitive functions, but the infrastructure required to support the new digital economy creates a much wider labour chain across construction, engineering, grids, generation, maintenance, cooling, power electronics, BESS, telecommunications, security, industrial software, data centres, system operations, cybersecurity, transformer manufacturing, cables, substations, equipment installation, water treatment, technical services and infrastructure finance.

The completed data centre may require relatively few permanent employees compared with its total capital investment, but the energy infrastructure feeding it mobilises projects for decades. A new substation requires design, construction and maintenance. A transmission line requires civil engineering, materials and operation. A solar park requires installation and O&M. A nuclear station needs thousands of workers during construction and a specialised technical workforce for decades. A BESS requires integration, commissioning, EMS, SCADA, safety systems and maintenance. A more digital grid needs analysts, operators and cybersecurity specialists. The new employment economy therefore does not sit only inside the building containing servers but throughout the ecosystem allowing those servers to continue operating.

This creates an exceptional opportunity for industrial regions that understand they do not need to compete directly with Silicon Valley in model design to capture part of the wealth created by artificial intelligence. They can compete in electricity, infrastructure, maintenance, engineering, land, cooling, storage, manufacturing, ports, logistics and technical education.

Latin America possesses renewable resources, gas, hydro potential, minerals, extensive land and ports. Europe possesses interconnected grids, industrial capacity, financial markets, nuclear expertise, renewable resources and digital demand. The Middle East possesses capital, energy, land and technological ambitions. The United States combines technology, capital, gas, renewables, nuclear power and a huge market. Asia possesses manufacturing chains and electrical expansion at extraordinary scale. The artificial-intelligence race therefore becomes a race to construct complete energy ecosystems rather than merely to train models.

Finance immediately enters the centre of the challenge because the required investments are enormous and cannot rely exclusively on technology-company balance sheets. Generation, transmission, storage, data centres, substations and digital networks require long-term capital, debt, equity, bonds, infrastructure funds, power-purchase agreements, capacity contracts and hybrid structures. This opens a new chapter for bond markets because AI expansion can create a wave of issuance intended to finance electrical and digital infrastructure backed by structural demand.

A data centre with long-duration contracts, secured electricity and investment-grade customers can form the basis of highly financeable infrastructure. A transmission line unlocking several gigawatts of demand creates economic value beyond its regulated tariff because it enables additional investments. A BESS located at a constrained node can generate multiple revenue streams while deferring immediate network reinforcement. A generation asset capable of delivering firm capacity for decades can become extraordinarily valuable to digital clients prepared to pay for energy security.

Institutional capital will increasingly examine this architecture because the infrastructure supporting artificial intelligence combines secular demand, long-term contracts, physical assets and enormous financing requirements. BlackRock has identified energy systems and land permitting as potential constraints on the AI buildout and has highlighted the role private capital can play in financing future energy projects, which matters because it demonstrates that markets are beginning to understand the next phase of AI will not be financed only through technology equities but through construction of the physical systems those companies require.

Goldman Sachs can analyse the transformation at the intersection of capital markets, utilities, infrastructure and technology. CPP Investments represents the kind of long-duration institutional investor capable of considering grids, energy and digital assets within infrastructure portfolios. Galdana Ventures fits within the logic of technology and innovation capital. BalGreen’s opportunity is not to imply relationships that do not exist but to construct projects with the scale, data quality, governance and execution capacity required for capital of that level to understand and potentially finance them.

The essential point is that electricity required by artificial intelligence should stop being treated merely as an operating expense and begin being structured as an asset platform. If a data centre needs five hundred megawatts, the question should not simply be where to buy electricity but which portfolio of generation, storage, grid capacity and flexibility can guarantee those five hundred megawatts with the lowest possible risk over fifteen or twenty years. Part may come from renewables through long-term contracts, another part from nuclear or other firm generation, another from storage, another from demand flexibility and another from dispatchable backup.

The data centre itself can become a flexible asset if selected workloads can be shifted temporally or geographically according to grid conditions, and this innovation is particularly powerful because it transforms a large consumer traditionally treated as inflexible load into an active participant in the electricity system. Computation can be scheduled. Certain training processes can be moved toward hours of greater energy availability. Lower-priority loads can be reduced during peaks. Facilities located in different regions can redistribute selected workloads. Servers and cooling systems can respond to price and congestion signals. Artificial intelligence can therefore use its own optimisation capacity to reduce the energy problem it is creating.

This will become one of the most important innovations of the next cycle: grid-interactive data centres capable of partially adapting electricity demand to system conditions without compromising priority services. The economic value can be enormous because reducing a few percentage points of load during critical hours can avoid or defer expensive generation and transmission investment, lower capacity charges and accelerate grid connection. Artificial intelligence then stops being only a passive load and begins to become a source of flexibility.

The opportunity extends to ports because many possess industrial land, electrical connections, fibre infrastructure, access to water, logistics capabilities and proximity to large metropolitan areas and can become strategic locations for digital and energy infrastructure. BalGreen Ports can view a port not merely as a maritime terminal but as a platform where BESS, generation, shore power, data centres, equipment assembly, component storage, hydrogen where economically rational, heavy-vehicle charging and grid services converge. A port with abundant and controlled electricity can attract industry, processing, cold storage, advanced logistics and computing. Energy stops being an auxiliary service and becomes a core product of port competitiveness.

This evolution also requires reconsidering the relationship between data centres and communities because building enormous facilities without explaining who finances the grids, which jobs they create, how much water they use, what energy they consume and how they contribute to the territory can generate social opposition. The licence to operate must be built around measurable benefits.

A project can finance a new substation that also benefits local industries, invest in renewables, use BESS to reduce pressure during critical periods, reuse heat, finance technical training, pay taxes, create suppliers and contribute to grid modernisation. The difference between a data centre perceived as a parasitic load and one perceived as economic infrastructure depends on how value is distributed.

BalGreen can intervene precisely inside this architecture through a proposition integrating energy, data, efficiency, storage, infrastructure and finance. DOIX can measure demand, load curves, peaks, availability, losses, cooling consumption, energy performance, emissions, BESS condition, grid capacity and savings produced by interventions.

BalGreen can design the operating architecture by selecting where to integrate storage, how to reduce peaks, which processes can become flexible, which generation should be contracted, how microgrids can be used, how waste heat can be recovered, how cooling systems can be improved and how port or industrial infrastructure can be integrated. DOIX can then verify the outcome and demonstrate how much consumption was reduced, how much peak demand was avoided, how much operating time was protected, which emissions declined and how much additional capacity was released.

That evidence can support financial structures, bonds, infrastructure debt, performance-linked contracts or asset portfolios intended for institutional capital. The doctrine remains consistent and becomes more powerful as the asset grows: measure before investing, identify the loss, design the correct solution, implement, verify and finance the performance.

The risk in this new economy is that governments and companies repeat the mistake of pursuing capacity before efficiency. Building new generation is necessary, but wasting expensive electricity inside poorly cooled data centres is irrational. Reinforcing networks is necessary, but ignoring available load flexibility can result in oversized infrastructure. Installing batteries can create value, but doing so without measuring nodes, demand and congestion destroys returns. Building data centres is strategically important, but locating them where water, electricity or the grid are already under severe stress can create conflict and raise costs. Artificial intelligence demands a systems approach because its scale can turn small design mistakes into enormous losses.

Financial risks are equally important. Expectations around AI can stimulate construction based on overly optimistic demand forecasts, create overcapacity in certain regions or push land, energy and equipment prices to levels difficult to justify. Investors must distinguish infrastructure supported by strong contracts from purely speculative projects. Banks must analyse technology obsolescence, customer concentration, electricity availability, interconnection timelines, supply contracts and residual equipment value.

Utilities must avoid allowing a single customer to represent an excessive share of regional demand without adequate guarantees. Governments must determine who finances required network expansion and protect residential customers from unjustified cross-subsidies. The AI energy race can produce extraordinary growth, but only if infrastructure costs are allocated correctly and the system avoids privatising returns while socialising risks.

The coming years will show increasing competition between territories to offer firm electricity to digital industry, long-term generation contracts between technology companies and power producers, restoration of nuclear assets, new nuclear construction where economics and policy support it, expansion of flexible gas capacity where reliability requires it, extraordinary renewable and BESS growth, larger investment in grids, forward procurement of transformers, microgrid development, data centres capable of partially modifying their demand and new financial instruments connecting institutional capital with digital-energy infrastructure.

The centre of gravity of artificial intelligence will therefore shift from a purely technological narrative toward an industrial architecture. Chips will remain fundamental, but without electrons there is no artificial intelligence. Models will continue improving, but without grids they cannot scale. Agents will automate processes, but without data centres they cannot operate. Software will continue moving at extraordinary speed, but physics will retain construction timelines, permitting requirements and material constraints.

Electricity becomes the new industrial currency because it determines who can convert digital capability into real economic activity. The company controlling energy controls cost. The city possessing grid capacity attracts investment. The port combining electricity, storage and data can attract new industries. The utility capable of anticipating demand captures value. The fund financing the correct bottlenecks can generate returns for decades. The region training technicians can create durable employment. Artificial intelligence does not eliminate the physical economy. It makes it more important.

Which countries actually know how much new electricity demand their grids can absorb before investment begins to stall, which governments are training electricians, engineers and technicians with the same urgency with which they discuss programmers, which utilities will turn data centres into flexible resources rather than treating them only as loads, which technology companies will finance generation and grids rather than transferring the entire cost to the system, which banks and funds will recognise that the real asset behind artificial intelligence is energy infrastructure, which ports can become platforms for both electricity and computing and how much value can BalGreen create if DOIX measures demand and losses, BalGreen designs the energy and operating architecture, implementation reduces costs and releases capacity, DOIX verifies performance and global capital transforms that measurable infrastructure into a new industrial asset class?

illuminem Voices is a democratic space presenting the opinions of leading Sustainability Thought Leaders, their views do not necessarily represent those of illuminem.

The world needs sustainability knowledge. At illuminem, no interest group or shareholder can influence our work. Thank you for supporting our mission to make high-quality and independent sustainability information free for all. Every contribution helps. Thank you for donating today.