AI Pushes Datacenter Electricity Consumption to 1,050 TWh by 2030 — Who Will Pay the Bill?

The explosion of generative artificial intelligence is transforming the global energy landscape. By 2030, AI datacenters will consume 1,050 terawatt-hours per year — more than India or Japan. This shift creates a datacenter energy market that will grow from $12.5 billion to $70.6 billion by 2035. But who will pay this bill? And where do investment opportunities hide? FINXIA Capital presents a comprehensive economic overview.

The projections are staggering. According to the latest Electric Power Research Institute (EPRI) study, global datacenter consumption could triple by 2030 to reach 1,050 TWh. This figure represents 4% of current global electricity production. The generative AI industry alone — ChatGPT, Claude, Gemini, and their successors — would be responsible for 60% of this growth.

The datacenter-dedicated energy market is experiencing unprecedented acceleration. From $12.5 billion in 2025, it should reach $35 billion by 2030, then $70.6 billion by 2035 according to BloombergNEF. This 460% growth over ten years places datacenter energy among the most dynamic segments of the global economy, ahead of traditional renewable energies.

Consequences on National Energy Markets

In the United States, datacenters will represent 8% of total electricity consumption by 2030, compared to 3% today. In Virginia, the US state hosting the largest concentration of datacenters in the world, this share will already reach 25% by 2027. Regulators are alarmed: demand is growing faster than production capacity, threatening grid stability.

In Europe, the situation is even tighter. Ireland, which hosts European datacenters for Microsoft, Google, and Amazon, has seen their consumption exceed that of all households in the country combined. France, with 18 GW of demand in queue, risks seeing its industrial electricity prices increase by 25% to 35% by 2028 if investments in production and transmission do not keep pace.

Who Will Pay the Bill? The Economic and Regulatory Debate

The financing question divides stakeholders. Hyperscalers — Microsoft, Amazon, Google, Meta — negotiate long-term Power Purchase Agreements (PPA) to secure their supplies. These contracts, signed for 10 to 20 years, represent colossal financial commitments. Microsoft has already contracted more than 20 GW of renewable capacity worldwide.

But these PPAs don't solve everything. They shift price risk to energy producers, who demand increasing risk premiums. The average cost of a solar PPA for datacenters increased by 18% in 2025. Meanwhile, European regulators are considering specific taxes on datacenter consumption to finance grid infrastructure, following Singapore's lead with its sectoral carbon tax.

FINXIA Capital's Investment Thesis: Energy as Alpha

FINXIA Capital considers energy to have become the most important strategic variable in datacenter investment. A site capable of generating or storing its own energy — through solar panels, batteries, or SMR contracts — benefits from a structural competitive advantage translating into 20% to 30% lower operating costs.

Our analysis shows that datacenters integrating a proactive energy strategy achieve an IRR of 18% to 22%, compared to 12% to 14% for sites dependent on the national grid. This 6 to 8 percentage point gap primarily stems from energy cost control and the ability to negotiate more attractive lease contracts with hyperscalers, who value energy resilience.

What Investors Must Understand

The energy bill of AI datacenters is not a marginal P&L line item: it is the factor that will determine which assets survive and which disappear. By 2030, energy will represent 60% to 70% of an AI datacenter's operating costs, compared to 40% today. Sites without a clear energy strategy will be the first victims of this evolution.

For the investor, the opportunity is twofold. First, invest in datacenters with partial or total energy autonomy. Second, participate in financing dedicated production and storage infrastructure. The $70.6 billion market by 2035 offers attractive returns at every link in the chain. This is the energy conviction guiding FINXIA Capital in its TITAN DC AI deployment.