A propitious week, perhaps, has just passed. On Tuesday, the Trump administration announced a $500 billion plan by Softbank, OpenAI, and Oracle – or at least a plan by their CEOs – to build data centers in the United States. As the week unfolded, a multi-GW facility was announced for India, another billion was announced to support OpenAI rival Anthropic, and visions of nuclear power were in the air.

The current interest in AI-Everything is driving all of this enthusiasm (or perhaps FOMO). Global angst over finding enough electricity to power all the new data center footprint purportedly stomping in soon. The $500B commitment would build about 40GW of new data centers in the US, which is more than twice the entire existing footprint. It's good to remember that Northern Virginia's Data Center Alley, by far the world's most dominant region and the model for everywhere else, has not yet reached 4GW and has already strained its local utility beyond its limits.

Similar anxiety is rife in Singapore, which feeds more than 12 percent of its electricity to data Centers, as well as in Denmark, Ireland, and the Netherlands, each of which have data center footprints that consume more than 5 percent of their grids. Several other EU nations, including Germany, are now over 3 percent.

Looking at the world as a whole shows a different picture. IDCA Research finds that data centers currently consume 1.1 percent of the world's electricity, with 2.8 percent of the US grid supporting almost half of the world's data centers.

And It's also worthy to note that most of the developing world consumes a tiny bit of tiny local electricity grids. The 80 or so countries of the developing world (not including India) produce less than 20 percent of the per-person electricity of the developed-world standard, yet their data center footprints consume less than 0.1 percent of their underdeveloped grids. There is more data center footprint in a medium-size facility in the US than in the developing world in total.

Meanwhile, India's current 1GW data center footprint consumes about 0.5 percent of its struggling grid, while even China's growing, multi-GW footprint consumes less than 0.5 percent of its massive electricity output.

Yet even with this imbalance, anxiety persists in the developed world. Nuclear energy has emerged as part of the energy-shortage solution, a panacea for names as influential as Bill Gates and Jeff Bezos. It seems it will sink in to them (and many others) soon enough that new nuclear energy in any guise faces long, steep regulatory and societal hills before coming on line. Elsewhere, with sustainability now under siege, renewables are gaining more production share by the day, yet still exist in a world in which 70 percent of the electricity produced is not renewable.

Back to the developing world, the data makes it very clear that these nations not equipped to handle a large influx of investment and construction of the massive new AI centers being touted by the big players. Should these gigawatt/gigadollar visions come to even partial fruition in the developed world only, it also becomes very clear that the world's income imbalance and disparity will only become more profound. This is an ongoing, unpropitious issue that seems to be of little to no interest to the movers and shakers within the AI-Everything movement at the moment. AI-Everything is about power, electrical and political.