X uncovers a Chinese bot farm amplifying fear of AI data centers in the US

🕒 Published on Zendoric: September 1, 2026 · 00:48
✨ AI-generated · how it's made
X has announced the identification of a network of roughly 200,000 accounts linked to Chinese operators. Within that massive network, the company points to a much smaller subgroup — around 200 accounts — dedicated specifically to posting content opposing artificial intelligence data centers in…
X has announced the identification of a network of roughly 200,000 accounts linked to Chinese operators. Within that massive network, the company points to a far smaller subgroup —around 200 accounts— dedicated specifically to posting content opposing artificial intelligence data centers in the United States. According to X's Global Government Affairs team, these accounts operated "in a manner that could manipulate a legitimate debate about American energy and AI policy," combining text posts with AI-generated cartoons portraying the operators of these centers enriching themselves at the public's expense.
The content from those 200 accounts centered on two specific claims: that the construction of AI data centers is driving up household electricity prices and that it is putting excessive strain on the power grid. Notably, at least two of the images used as examples had already appeared in a similar report published by OpenAI in June, which documented how ChatGPT users based in China were asking the chatbot how to influence Americans' opinion about these facilities. That report generated posts with the same message about rising electricity costs.
The most striking aspect of the case, and the point the article itself underlines, is that those "claims" attributed to the disinformation campaign are not actually false. Bloomberg has documented electricity price increases of up to 267% in areas near data centers, and Goldman Sachs has calculated a 6.9% year-on-year rise in electricity prices through February 2026, with expectations that they will keep climbing as more facilities are built. In other words, a foreign propaganda operation is amplifying a problem already verified by independent financial and journalistic sources, not fabricating one that does not exist.
X has not disclosed what content the network's remaining 198,800 accounts were posting. The article suggests, based on the precedent of OpenAI's June report, that they were probably devoted to criticizing US tariffs and technology policies, though this remains a reasonable assumption rather than a fact confirmed by X.
The backdrop to this story is the growing public pushback against the expansion of data centers in the United States, a phenomenon that already had momentum of its own before there was any talk of Chinese interference. That pushback has triggered a political counterattack: outlets such as Fox News have compared protesting against these facilities to opposing tank and aircraft factories during World War II, and President Donald Trump himself has declared that those who oppose the construction of new data centers "are making a mistake."
The reasons for public discontent are well documented and do not depend on any bot campaign. These facilities' electricity consumption is extraordinary: a campus planned in Michigan has an estimated demand of 1.4 gigawatts, comparable to the output of a nuclear plant, on top of generating complaints from neighbors about traffic and the loss of farmland. Water is another flashpoint: data centers in Virginia consumed at least 1.85 billion gallons in 2023, up from 1.13 billion in 2019, and in Georgia one development used 29 million gallons before it even received a bill, while nearby residents reported drops in water pressure. Add to that the constant noise and the additional pollution from the use of diesel generators and gas turbines when the grid cannot keep up.
The conclusion that emerges from the article is more ironic than alarming: China may be exploiting this controversy for its own geopolitical ends, but it did not invent it. Opposition to AI data centers in the US was already broad, intense and grounded in verifiable facts —rising electricity prices, strained grids, runaway water use— before any automated account appeared. That leaves an open question about the real usefulness of this kind of influence operation when the target audience already shares the conclusion it is meant to be "taught."
🔗 Related on Zendoric
- Monday.com lays off 20% of its workforce (around 630 employees) to refocus the company on AI · 2026-07-24
- IBM and the sub-1 nanometer frontier: the silicon race reorders itself around AI · 2026-06-27
- Hartford bets on teaching AI to its young people before the job market leaves them behind · 2026-07-01


