Facebook (FB) is about to make another business blunder. Nuclear power is the most expensive and slowest to build form of energy. It cannot just be built anywhere due to large water requirements and gets major NIMBY pushback wherever its proposed.
Even utility-scale nuclear powerplants in the US (and around the world) built by well-established experience nuclear builders frequently end up over-budget by billions and over-schedule by years. It's the longest running joke in the energy industry.
FB is proposing 1GW to 4GW sized power plant which puts it in the utility-scale nuclear power plant. But look at the latest US nuclear reactors and power plants of this scale:
(1) Georgia Vogtle: Overbudget more than double from original $14B to $35B. Over-schedule 7 years with total construction time of ~11 years for each reactor.
(2) France Flamanville-3: Overbudget triple from original €3.3B to €9.9B. Over-schedule 12 years with total construction time 17 years. Launched recently but only running in reduced power output for about 1 year for additional testing.
FB much better off using the large datacenter roof areas, parking lot areas, and adjacent lands building photovoltaics, local battery storage, and buying renewable energy from the grid thus supporting the expansion of grid renewable energy.
US has huge solar PV growth. The US is on track to produce 218GWh of actual solar generation in 2024 an increase of 25% (54.5GWh) over 2023. Solar actual WattHour generation is growing at some 30% compound annual growth (CAGR) over the past 8 years or so. That means a doubling of generation every 2.6-years and 10x every 8.8-years.