July 8, 2026
Gridlocked and giga-mad
What's slowing down the AI buildout
AI’s not running out of hype — it’s stuck waiting for the power hookup wars
TLDR: The big AI buildout isn’t mainly short on electricity — it’s stuck waiting to get connected to the power system. In the comments, that sparked a messy fight over clean energy, nuclear power, off-grid workarounds, and whether AI demand is even real enough to justify the frenzy.
America’s giant AI boom has found a very unglamorous villain: the power grid paperwork nightmare. The article says the US likely has enough electricity overall, but massive new AI data centers — including OpenAI and Softbank’s eye-watering $40 billion Stargate project in Texas — can’t just plug in and go. These sites need huge amounts of power, and the real delay is getting that electricity delivered to the right place and approved through a slow, jammed-up connection process.
But in the comments, readers turned this into a full-on energy cage match. One camp was furious that clean energy keeps getting dragged into politics, basically arguing that the country is making this harder than it needs to be. Another group came in with the classic "we should’ve built more nuclear" rallying cry, mocking the idea that “windmills and sweaters” are enough for an AI future packed with robots. Meanwhile, skeptics were side-eyeing the whole premise, asking whether AI growth is really being held back by electricity at all — or whether the industry is just buying hardware faster than reality can support.
And then came the survivalist-tech bro energy: one commenter pointed to Crusoe skipping the grid by using old batteries and wasted energy, which sounds either brilliantly scrappy or like the plot of a sci-fi side quest. The overall mood? Part panic, part policy rant, part meme. Everyone agrees AI wants an absurd amount of power — they just can’t agree on whether the answer is solar panels, nuclear plants, off-grid hacks, or a giant reality check.
Key Points
- •The article says the flagship Stargate AI computing campus in Abilene, Texas is expected to cost more than $40 billion and is tied to a broader project led by OpenAI and SoftBank.
- •Stargate is expected to draw 1.2 gigawatts at peak load, which the article compares to the electricity use of about 313,000 median American family homes.
- •A cited report projects that global AI computing power could reach 100 gigawatts in 2030 if 2025 growth rates continue.
- •The article argues that the primary bottleneck to AI infrastructure growth is not an overall lack of electricity but delays in connecting projects to the power grid.
- •Before new infrastructure can connect, grid operators must study impacts on power flows and needed upgrades, and the article says this interconnection process is heavily backlogged.