AI as impetus for a dramatic increase in datacenters, and associated electricity demand, headlines almost every major energy event right now.
Common framing:
(a) Demand: Datacenters require massive amounts of electricity, owners prioritize reliability and are very price elastic, and the value of datacenters is accelerating due to the pivotal role data plays fueling AI.
(b) Supply: Wires transmitting power are already stretched to their limits and take 3-5 years to permit and build, and stakeholders who spur such action are slow and bureaucratic.
Scary scenario: Factoring (a) and (b) together, electricity prices could shoot up due to demand dramatically exceeding supply. In places where power is $0.08/kWh nowadays it could increase 5x to $0.40/kWh.
Key factors more folks should understand:
(a) Electricity rates are sector-specific (residential, commercial, industrial), and they often use tiered pricing. In the medium-term, if datacenters become outlier electricity consumers, they're likely to receive specialized, expensive electricity pricing from the grid than households.
(b) Many but not all datacenter players are savvy about off-grid power procurement. Hyperscalers like Google, Microsoft, Amazon, Apple, and Meta Facebook have been building wind and solar plants to power their datacenters for over a decade.
(c) Fossil-based backup generators ensure reliability. Just last week, a datacenter leader at Microsoft posted: "Gas is much cleaner and lower in cost than diesel... By pulling together the utility, our world-class Microsoft Energy Team, and gas generator manufacturers, we were able to collaborate and engineer a solution that has worked very well for many years and serves as a model for how datacenters can be an asset to the grid, not a liability." This was the first time in 10+ years I can remember Microsoft bragging about fossil fuel use and painting gas as "clean" just because it's not diesel.
Winners from these dynamics:
(1) Technologies that squeeze the most power possible out of existing power lines. Such technologies include line monitoring (ex. LineVision), virtual power plants, and demand response.
(2) When grid power price is higher, economics become much more favorable for behind the meter generation at all scales (residential, commercial, and utility). Solutions from residential, community, commercial, and utility-scale solar to behind-the-meter wind farms should benefit.
(3) In reality, fossil fuel (gas and diesel) backup generators will likely see very strong demand, considering even Microsoft, an admired stalwart in corporate sustainability, is apparently not embarrassed to greenwash natural gas as clean in this context. Ideally, though, decarbonized gensets like that from ClearFlame Engine Technologies will see massive adoption, and datacenters will become increasingly co-located with with decarbonized baseload power such as geothermal and nuclear.
#CleanEnergy #RenewableEnergy #AI #Datacenters
Accelerating our Energy Transition
1moThanks for hosting Peter Sopher! Conversation was great, and that was hands-down the biggest crème brulée I've ever had... Who serves it in a bowl?? I guess things really are bigger in TX.