Google (NASDAQ: GOOGL) has just put an enormous $25 billion down on the future of data, energy, and artificial intelligence in the United States, pointing its latest investment straight at the sprawling PJM Interconnection grid. This power move comes as part of an aggressive expansion strategy, blending data center construction with equally ambitious clean energy procurement, and highlighting the intense power hunger driving AI’s meteoric rise across the country.
The PJM grid probably isn’t the first thing that comes to mind when you think of the forces shaping artificial intelligence, but it is the country’s largest electric network, covering 13 states from the mid-Atlantic all the way into the Midwest, and touching parts of the South. With more than 65 million end users and 185 gigawatts of generating capacity, PJM operates the country’s most complex and significant energy backbone, balancing the electricity needs of 1,400 generation units across a landscape stretching from Chicago to New Jersey.
Google’s $25 billion investment will unfold over the next two years, dedicated to building new data centers and AI infrastructure across the PJM footprint. This announcement comes as the company and its tech peers crowd the market for energy, a race accelerating as AI workloads explode in scale. In practical terms, Google’s expansion isn’t just about laying more fiber and building warehouses filled with servers; it’s about locking down enough reliable power to fuel these new digital factories, knowing the AI boom is only intensifying that demand.
Recently, Google’s data centers have been using stunning volumes of electricity. In 2024 alone, the company consumed 30.8 million megawatt-hours (MWh) of electricity globally just for its data centers, up from 14.4 million MWh in 2020. That represents more than 95% of the entire company’s electricity use. In the United States, data centers at large consumed about 4.4% of all electricity in 2023, with projections showing that could triple by the end of the decade as AI becomes more deeply embedded in our lives and businesses. AI workloads, and especially large models like those powering voice recognition, translation, or image generation, are exponentially more energy-intensive than conventional web hosting or search. A single interaction with a generative AI system can use up as much as ten times the electricity of a regular Google search.
All told, Google’s current data center fleet by itself accounts for around 7 to 10 percent of the total energy consumed by all data centers globally, which hints at the scale of operations required to keep the gears of AI turning. These are not the silent, out-of-sight operations of old. Instead, data centers have become critical, cities-within-cities where energy, cooling, and compute run around the clock and get smarter and more demanding each year.
With the AI revolution showing no signs of slowing, Google is racing to find energy solutions that can match both its scale and its climate ambitions. In this context, the new partnership with Brookfield Asset Management stands out as a critical support beam for Google’s plans. The agreement will see Google secure up to 3 gigawatts of hydroelectric capacity in the US, primarily drawn from upgraded and modernized power facilities in Pennsylvania, including the Holtwood and Safe Harbor hydro plants. These twenty-year power purchase deals represent about $3 billion and promise to feed Google’s new PJM region data centers with long-term, carbon-free electricity.
Google’s approach is a clear response to dual pressures, the rapid growth of artificial intelligence and the growing urgency to curb emissions, even as global data center fuel use climbs year by year. The Brookfield deal helps Google take a significant bite out of its carbon footprint in one of the world’s most energy-hungry markets, all while keeping its AI ambitions on track.
Artificial intelligence isn’t just a clever software upgrade, it’s a new kind of energy-intensive industry. Looking forward, if current forecasts pan out, US data centers could end up using as much as one-fifth of the nation’s total electricity within a decade. A significant share of that increase will be laid at AI’s feet. The vast, interconnected web of servers and cooling systems enabling today’s voice assistants, translation apps, and search engines are drawing more power each year, and there is little sign of that slowing.
For Google, that means the success of its $25 billion bet doesn’t just hinge on building faster servers or smarter AI. It depends on whether the largest electric grid in the country, with an influx of clean energy from partners like Brookfield, can keep pace with the demands of tomorrow’s digital economy.
