Bill Gates's recent declared that he has changed his mind and no longer views climate change as a serious threat to the future of humanity is not just a major blow to the "climate crisis" camp but likely signals the beginning-of-the-end of the great climate scare.

Gates wasn't the first to suffer a massive rethink on climate. Recently The Sierra Club, so long a champion of the we're-all-gunna-die crowd, has changed its mind and has declared it will start emphasising "social justice" rather than the environment. 350.org, previously active in opposing CO2 emissions, has closed due to insufficient funding.

The recent COP30 gabfest was a disaster for the CO2 fetishists. No leader of any major economic power bothered to attend. And precisely nothing was. Europeans were asking why Europe was destroying its economy when it only produces 6% of the emissions.

Why the sudden Road to Damascus enlightenment? Two words..."Artificial Intelligence".

AI is a massive user of electricity That's why companies like Google and Microsoft are partnering with uranium companies to power their AI farms. It’s calculated that the USA will need to build 100 1 gigawatt power stations per year. Renewables can't do it.

Any country that has ambitions of being part of the AI revolution has no choice but to massively increase their electricity production of reliable 24 hr output. AI can’t just be run when the sun’s out.

It’s always been clear that the renewables push involved a massive transfer of wealth to the investor classes via the subsidies that they convinced governments to pay. But now there’s a different vehicle for their money and they are abandoning renewables and going all in on AI farms and the electricity infrastructure they require.

Gates realised that his push for renewables was incompatible with Microsoft’s desire to be a leader in AI. Nothing more noble than that.

Our government tells us we need to join the renewables mania to remain competitive. But being competitive means joining the biggest thing since the industrial revolution. Hopefully our leaders will wake up to that before the world passes us by.

Apparently small modular reactors are better suited to the needs of mining and mineral processing than AI. Gate's sodium cooled reactor has molten salt storage to better allow a variable power output. Amazon is partnering with X-energy's pebble bed smr. Copenhagen atomics is using molten salt with a thorium blanket to breed fuel, although the process takes twenty years to double the fuel. There is one company in Texas using wind energy for a data centre, but I don't think this sort of example common.

https://www.texastribune.org/2025/10/21/south-texas-data-center-renewable-energy/

I wish these companies every success and am grateful for the support Trump is giving to the nuclear industry. Successful smrs will bring about the demise of wind and solar grifters. Although the koalas might not be threatened, it is still horrible that they are losing their lives because of those horrible grifting con artists.

I wonder how many here remember playing "xbill"?

- http://www.youtube.com/watch?v=pnTqlysDaGk

Artificial intelligence cannot operate effectively on renewable energy alone due to the fundamental incompatibility between the requirements of AI computation and the inherent limitations of renewable power sources.

Large-scale AI systems, such as training frontier models and providing continuous inference, demand vast, uninterrupted electricity. Training a single large language model consumes thousands of megawatt-days of power, while serving millions of users requires sustained, always-on capacity from data centres filled with tens of thousands of GPUs. These workloads cannot tolerate frequent interruptions, as latency-sensitive applications demand reliable, continuous performance.

Renewables, however, are intermittent and non-dispatchable. Solar generation is limited to roughly 25% of the day and is highly sensitive to weather, while wind output fluctuates unpredictably with localized conditions. Even with geographic diversification, renewables produce electricity less than 30% of the time, resulting in multi-day periods of near-zero output, such as wind droughts or solar dark periods. No combination of solar and wind can guarantee the continuous power that AI infrastructure requires.

Battery storage, the primary proposed solution for intermittency, is inadequate for this challenge. Utility-scale lithium-ion batteries provide discharge durations of only 4–10 hours, far short of the multi-day storage needed to bridge extended renewable lulls. Supplying hundreds of gigawatt-hours to sustain a large data centre for three days would require battery capacity exceeding current global annual production by orders of magnitude. Alternative storage options, such as flow batteries or pumped hydro, are either insufficiently scaled or geographically limited.

Consequently, renewable energy cannot independently support the continuous, high-density power demands of AI. Intermittency creates persistent gaps that neither current nor foreseeable storage solutions can fully resolve. Reliable AI operation requires dispatchable, low-variability sources—such as nuclear power, which achieves capacity factors above 90%—to provide the stable baseload electricity that intermittent renewables alone cannot deliver. Without such supplementation, a renewables-only approach is physically incapable of meeting the unrelenting energy needs of advanced AI systems.