Can anyone really comprehend what 128 GW of new demand in 5 years actually looks like? Is this even possible?
These are the forecasts from
@RobGramlichDC and co at grid strategies which are derived from utility forecasts submitted to FERC.
Most respond with skepticism towards the demand side. But colleagues and I at CSIS are tracking the chip orders, actual datacenter FIDs, and scaling laws... these numbers are in the right ballpark. It's not for nothing that every major hyperscaler has reversed decades of no-nuclear energy procurement strategies. The public is just catching up to what the hyperscalers recognized ~24 months ago.
The best reason to be skeptical of this forecast is because of Supply Side constraints.... can we even build the generation and transmission infrastructure to serve this demand at this speed? Demand will only grow as fast as our grid can reliably service it.
Stakes are high: every GW of demand not-served because of supply side constraints is lost economic value, jobs, tax-base and has profound strategic implications for global US leadership on AI.
Major work to be done in DC on so many long-standing policy issues: interconnection queues, stalled Tx expansion, gas-electric reliability, seams management, and new nuclear construction. The basic political terrain and policy paradigm for electric-power sector of last 30 years is now hopelessly out of date.