Big Tech including Amazon, Microsoft, Alphabet, and Meta have collectively committed around $635 billion in capital expenditure for 2026, with the bulk of that going toward AI data centers.
The scale of this investment is historically unprecedented. But as billions pour into chips, servers, and infrastructure, a more pressing question is surfacing – where is the power to run all of it actually going to come from?
S&P Global estimates that the leading U.S. tech platforms had mapped out around $635 billion of combined spending on data centers, chips and other AI infrastructure for 2026, up from $383 billion in 2025. In fact, these companies are now raising tens of billions in bonds to fund this buildout, given that the speed and scale of AI infrastructure has outpaced their operating cash flows.
Alphabet, for instance, recently raised roughly $20 billion in a single bond offering in February 2026 to fund its $185 billion AI infrastructure plan. And much of that money is going into building highly specialized AI data centers, packed with advanced chips from suppliers such as Nvidia.
However, S&P Global research leaders have warned that this spending boom now faces a major hurdle as energy and power costs rise, especially in the wake of the ongoing conflict between the U.S. and Iran.
Energy Costs And Grid Limits
Data centers have already caused an increase in the U.S. power demand. The International Energy Agency (IEA) has highlighted AI as one of the main new drivers of data center electricity demand worldwide, projecting that this worldwide demand is to quadruple by 2030.
The electricity demand that comes with data centers buildout is putting pressure on grids that were not designed to absorb it. In the U.S. alone, average electricity prices have gone up by around 13% since 2022.
As such, lawmakers and the U.S government have started questioning tech companies including Alphabet, Microsoft, Amazon and Meta over the grid constraints their growth is causing for utilities and citizens to handle.
How Tech Giants Are Responding
Faced with these pressures, Big Tech are beginning to adjust how and where they run their most energy‑intensive AI workloads. Some firms are exploring shifting processing to regions with more available capacity or cleaner energy, while others are turning more often to backup power and flexible demand programs so they draw less from the grid.
Google, for instance, has signed contracts with utility companies like AES Corp and Xcel Energy to secure dedicated, large-scale supplies of clean energy.
Chipmakers and AI companies are also exploring how power and energy can be harnessed in space. Nvidia-backed Starcloud recently raised $170 million in a Series A round to build orbital data centers that run on uninterrupted solar power in space.
What It Means For The AI Boom
For now, there is a wide acknowledgement of energy as a key risk factor that might affect the AI boom, especially given the dire situation going on in Iran.
Melissa Otto, head of research at S&P Global Visible Alpha, has warned that very high oil prices could force the hands of Big Tech to revise their AI spending in the first and second quarters of this year. This, Otto says, may bring a “really meaningful correction in all equity markets.”
What this means is that if the conflict between the U.S. and Iran keeps going, energy prices will remain high and grid constraints will worsen, which means parts of the $635 billion spending plan could be delayed, reshaped or diverted into efficiency and power projects instead of only AI compute expansion.
For Big Tech, the race to scale their AI ambitions has now become a test of who can secure the energy to match said ambitions.
