𝗔𝗜 𝗘𝗻𝗲𝗿𝗴𝘆 𝗖𝗿𝗶𝘀𝗶𝘀: 𝗦𝗼𝗹𝗮𝗿 𝘃𝘀. 𝗡𝘂𝗰𝗹𝗲𝗮𝗿
AI data centers need massive amounts of power.
The Colossus data center is expanding to 2 gigawatts. That matches the power use of 20 cities. OpenAI and Softbank's Stargate project targets 10 gigawatts. That equals 100 cities. The IEA expects data centers to use 945 terawatt-hours by 2030.
Companies must plan for these energy costs. Prof. Maximilian Fichtner provides a clear view on how to solve this.
Solar plus storage beats new nuclear power.
New nuclear plants are too expensive. Hinkley Point C costs over 54 billion pounds. Some new plants deliver power at double the market price. Small Modular Reactors are not ready for scale yet.
Solar and batteries work now.
The argument that solar needs constant sun is old. Large battery systems like the one in the UAE prove storage works. Storage near power sources reduces grid costs and handles peak loads.
Data center operators prefer renewables. A two-day weather report is easier to plan for than a massive nuclear plant that might fail.
The truth about energy prices:
- High electricity prices come from gas, not the exit from nuclear power.
- The EU Merit-Order principle means the most expensive plant sets the price.
- Gas prices drive the market.
- Hydrogen is currently too expensive at 16 to 18 euros per kilo. It needs to be 4 to 5 euros to compete.
Batteries beat hydrogen on price and availability.
If you scale AI, you must treat energy as a strategic tool. Solar and storage offer the best speed and price for the future.
Source: https://dev.to/everlast_ai/ki-energieproblem-prof-fichtner-uber-solar-und-speicher-1ofl
Optional learning community: https://t.me/GyaanSetuAi