The EU freaked out about EVs killing the power grid yet AI is doing the same thing much faster

I remember when the EU announced the combustion engine ban back in 2022, with the deadline set for 2035. The backlash was massive with endless statements and articles about how the grid can't handle it and that we don't have the infrastructure.

Fair concerns, honestly because Europe has seen some power outages since that announcement. But this has come to my mind (it was a slow work day today).

GPT dropped in 2018. By 2025, AI workloads aka the combined compute, storage and networking needed to train and run AI models (according to a quick google search) have already surpassed regular workloads. And that number is expected to nearly quadruple globally between 2025 and 2030.

For scale, a single AI query on something like ChatGPT uses around 0.0003 kWh this multiplied for each query each day. Gemini and others are similar. One AI server rack alone consumes the equivalent of 30 households (could be off, I've seen this on an article I can't seem to find anymore, there it was said that those racks are only consuming more and more energy and becoming more powerful). Even if that number is off, the general point stands.

According to this map (from reddit but I take it as fact for my laziness to properly research this), the EU currently has 28 AI data centers. That number is not going down.

The EU uses significantly less AI than the US. 43% of Americans use AI, while individual EU countries sit somewhere between 25–36%. Adjusting for population, I'd ballpark the EU consuming roughly half as much AI compute as the US — which I'm fully pulling out of my ass, but even at half, it's still a lot.

So here's the thing that doesn't add up to me.

The EU spent years arguing about whether the grid could handle EVs by 2035. Meanwhile, AI data centers have been ramping up power consumption this entire time, and the conversation is non existent if we take the scale into account? Either the grid is more resilient than the skeptics thought (I doubt that, like I mentioned there were some major power outages), or we've been adding massive load without really accounting for it, and the bigger picture is being completely ignored.

Supply wise, the EU has 125 nuclear reactors. The average age as of April 2021 was 33.4 years (yes that source has an agenda, grain of salt needs to be taken). Theoretical lifespans are 60–80 years, but in practice not a single plant has made it to 60. So a big chunk of existing capacity has a clock on it.

Currently there are 9 nuclear power plants being built. And a new nuclear plant takes an average of around 7 years to build (reddit comment I am taking as fact). 9 plants, for a fleet of 125 aging reactors, against a backdrop of exponentially rising AI power demand and an EV transition deadline of 2035.

With the renewables and nuclear (and exclusive to Germany coal power, which should actually be banned instead of nuclear, as long as nuclear is actually properly maintained) we are going to face plant (according to me).

I genuinely don't know what my point is here. I just find it kind of wild that EV induced blackouts were the headline for years, while AI quietly became a massive and growing power gobbler with way less scrutiny. (yes there are more and more articles but most are regarding water usage which also is a problem but you know, comparatively speaking, the power usage of AI is no where near as problematic in the mainstream)
I'm not saying we should put the breaks on AI but policies need to become realistic because with how we are moving forward now, is not sustainable, and I don't mean in a way how we weren't unsustainable til now but even worse)

Regardless of how you slice it with renewables, nuclear grid upgrades and expansions we're not building fast enough to keep up with what's coming. in my humble opinion which I formed in one afternoon.

Please discuss and tell me how wrong or right I am because I genuinely am curious on what is to come. I tried to think of many things but granted my form of research is a bit biased since I went in expecting this conclusion.