Artificial intelligence is often presented as a race for models. OpenAI, Anthropic, Google, Meta or Mistral AI compete with benchmarks, reasoning capabilities and billions of parameters. However, behind this competition there is a battle taking place neither in laboratories nor in research centers, but in electricity networks, transformer stations and the trading rooms of infrastructure funds.
The AI economy is entering a new phase, after the race for data, then for GPUs, now comes the race for gigawatts.
This development profoundly modifies the very nature of the sector; for more than twenty years, the digital economy thrived on an implicit hypothesis: electricity was available and abundant. Infrastructure was a secondary topic to software, AI reverses this logic. Every technological advancement requires more computing, more servers, and more energy. Limits are no longer solely algorithmic but become physical.
According to the International Energy Agency, data centers consumed around 460 TWh of electricity worldwide in 2024, this figure could exceed 1,000 TWh before the end of the decade under the influence of generative AI. For comparison, this would represent a consumption higher than that of Japan today.
This dynamic is already visible among hyperscalers, Meta plans to invest between $125 and $145 billion in its AI infrastructures in 2026. Microsoft could devote between $115 and $135 billion to its computing and cloud capabilities. Alphabet is pursuing a similar trajectory with an investment program of around $80 billion. Behind these amounts lie IT campuses whose energy needs now reach several hundred megawatts.
The Stargate project carried out in the United States illustrates this change in scale; its ambition is to develop several gigawatts of computing capacity dedicated to artificial intelligence. At this level, a data center becomes an energy consumer comparable to a large city or a major industrial complex.
This rise in power reveals a new form of scarcity. For the past two years, the industry has focused on advanced semiconductors and GPUs from NVIDIA. However, many players are gradually discovering that the real bottleneck is becoming access to energy.
If the capital is available, on the other hand, obtaining several hundred megawatts connected to the network within a time frame compatible with the ambitions of AI players becomes an increasing challenge. In some regions, connection times reach several years.
This reality is now appearing in regulatory debates; Mississippi has become one of the first laboratories of this new AI energy economy. The state has attracted several projects from AWS representing more than $13 billion in announced investments. To support this growth, the operator Entergy is developing several new production capacities, including three gas power plants representing a total of more than 2,200 MW of installed power. The estimated cost of this infrastructure exceeds $3.8 billion.
A report published by Synapse Energy Economics estimates that residential consumers have already contributed around $38 million to investments associated with the infrastructure intended to accommodate these new data centers, with an amount that could reach $74 million at the end of 2026. The authors especially highlight the impossibility of precisely verifying the distribution of costs due to the confidential nature of the contracts concluded between large consumers and the electricity operator.
Mississippi is not an isolated case; several American states are starting to create specific rate classes for very large electricity consumers. Virginia, Ohio, Kansas and even Pennsylvania are working on mechanisms imposing long-term commitments, financial guarantees or minimum consumption levels in order to prevent infrastructure costs from being transferred to other users of the network.
Europe has not yet opened this debate with the same intensity, yet the same tensions are gradually appearing.
France aims to become one of the main European AI hubs. Project announcements are increasing around Mistral AI, Data4, OpCore and even consortiums supported by international investors. The country has an obvious comparative advantage with largely carbon-free and relatively abundant electricity production thanks to nuclear power, but this abundance is less obvious when it comes to connecting several hundred megawatts in the same territory.
The issue is not limited to data centers, the automotive industry is demanding more electricity for its gigafactories, while hydrogen producers want to secure significant capacities. Decarbonizing heavy industry also requires massive electrification. Transport and residential uses follow the same trajectory. For the first time in several decades, several public policies are converging on the same resource: electricity…
The question then becomes less technological than industrial, because each gigawatt allocated to an AI campus is a gigawatt that will not be immediately available for other uses. The debate on financing is gradually hiding a new question: who should have priority in the allocation of electrical capacity?
This development is also transforming the role of hyperscalers. AWS, Microsoft, Google or Meta are no longer just technology companies. Their investment decisions now influence the energy strategies of entire regions. Like the large steel or automobile groups of the last century, they are becoming players capable of directing investments in networks and the economic development of territories.
Another player is emerging in this equation: infrastructure funds, Brookfield, BlackRock, KKR, Macquarie, Global Infrastructure Partners and even the Gulf sovereign funds are investing massively in data centers, electricity networks, energy infrastructure and production capacities. AI creates a new market for patient capital; the infrastructure necessary for its development will be amortized over twenty, thirty or forty years, while models evolve every six months.
This paradox is undoubtedly one of the most symptomatic of the current period, the fastest technological industry in history now depends on the slowest infrastructures to build.
The question “Who will pay for the gigawatts of artificial intelligence?” » therefore does not call for a single answer. Consumers, hyperscalers, network operators, infrastructure investors and governments will all participate, in one way or another, in financing this new layer of infrastructure.
The real question is who will decide their allocation? Which projects will be considered priorities? What place does Europe want to give to AI in relation to reindustrialization, transport or the energy transition?
For two years, the AI industry has been dominated by the battle of the models. The next decade could be dominated by infrastructure. And in this new economy, the decisive factor will perhaps no longer be the quality of the algorithms, but the capacity to sustainably mobilize gigawatts.
