Google no longer waits for overloaded utilities or distant grid upgrades to arrive. Alphabet is spending billions to bolt energy onto its AI infrastructure as surging AI power demand turns electricity into strategic leverage.
The acquisition of Intersect Power for $4.75 billion redraws how Google links energy projects to data centers. It gives Google campuses wired to solar and battery assets, a private power supply with debt-inclusive deal value that sidesteps transmission bottlenecks.
Why Google is moving off the grid for AI compute
Alphabet is paying 4.75 billion dollars in cash and assumed debt to buy Intersect Power so AI growth is not dictated by public utilities. The surge of generative‑AI traffic creates hyperscale compute loads that strain substations and raises grid congestion risk in regions where new lines are already delayed.
Google has spent years signing long‑dated contracts with generators rather than simply waiting for utilities to expand capacity. Deals with NextEra cover about 3.5 gigawatts of wind and solar, while TotalEnergies supplies 1.5 terawatt‑hours of contracted output that anchor a broader power procurement strategy. Intersect’s projects add on‑site generation to that mix, supporting long‑range reliability planning for AI campuses expected to draw hundreds of megawatts.
What Intersect brings to the table with power‑first data centers
Intersect Power has grown by pairing utility‑scale generation with large energy users on the same land. For Google, that approach translates into power-first data centers where dedicated gas, solar or wind assets feed high‑capacity connections straight into server halls. Alphabet can link new campuses to bespoke power plants, avoiding multi‑year waits for transmission upgrades and aligning economics directly with AI workloads.
Many Intersect projects add large lithium‑ion installations alongside wind and solar, turning intermittent output into something closer to a controllable plant. Those battery storage systems, often built with Tesla Megapacks, can firm supply for Google data centers and help bid flexible services into local power markets.
Grid congestion, local risk and the cost of concentrating megawatts
Concentrating hundreds of megawatts of IT load in one place turns a data facility into something closer to a power plant in grid studies. Regions that host several such sites quickly run into transmission bottlenecks as existing lines, transformers and substations hit their physical limits. Regulators respond by tightening capacity reserve requirements, forcing grid operators and utilities to procure extra generation simply to cover a handful of large AI campuses.
Such concentration also ties local economies to the fate of a few giant server halls, amplifying any disruption that hits them. Where grids are weak, megawatt-scale clusters of compute can trip offline during weather-related outages or cyber incidents, triggering cascading voltage swings that spill over into homes and nearby industries.
Climate and investor fallout as electricity becomes the constraint
Energy analysts now estimate that AI already consumes around 1.5 percent of global electricity, and the growth of large language models is still in its early stages. As demand jumps, operators lean harder on gas and coal units, creating carbon intensity spikes even while Google signs 1.5 terawatt‑hours of solar with TotalEnergies and explores advanced nuclear ventures such as Elementl Power to firm supply.
Investors now treat electricity as a determining input for AI platforms rather than a marginal utility charge. If cheap low‑carbon supply fails to materialise, models show sharp valuation sensitivity to power and greater stranded assets risk for data centers and chips, which Google seeks to counter through its 4.75‑billion‑dollar Intersect deal.