tyler fitch

@NiyerEnergy IMO they're just different configs. at same MW, twice the duration is twice the MWh. That'll be more expensive, but not twice as expensive. To me, your RT is more about what length of discharge is required to be in the money given system dynamics

"Long duration" batteries are literally worse! You have the same energy storage but it can charge and discharge less quickly? If that saves $$ (interconnection/equipment) and can have value for longer charge/discharge cycles that's great. But the meaningful unit here would be $/kWh storage, not "hours of charge". I think it's meaningful in the sense that batteries are cheap enough where this kind of thing can be built in the first place, which means its getting cheaper. This is true! But $/kWh is the unit that matters, and longer battery systems are simply less performant. Open to disagreement here tho

@kaLauntra @NatBullard we are so far past dignity at this point

@ty_fi @NatBullard LinkedIn Twitter cross posting?

Wow. Buckle up, folks -- the data center infrastructure boom is just getting started. h/t @natbullard and Halcyon! (part of me wonders, do we have the infrastructure to support this level of investment??? What amount of that capex has a reserved spot for power already?)

Like -- do these people have any self-respect? Good lord

IEA's press release from their Electricity 2026 report here: : iea.org/news/global-el…

Some helpful context from IEA via Semafor this morning: While data centers are shaping future load in the US, they are part of a bigger story of electrifying economies when we look across the world.

How many newsrooms could 1% of AI-related revenues sustain? 5%?

In the context of load that is grid-transformational and unprecedented, with emergent impacts from multiple projects -- what can we do to understand and quantify grid impacts? subscriber.politicopro.com/article/eenews…

Our tools for evaluating costs often assume a) changes at the margin and b) looking backward, rather than forward (as in cost-of-service-studies).

One thing I haven't seen discussed (and another insight from thinking about unexamined, load-bearing assumptions) -- do we have the tools and framework to understand the impacts/costs of new data centers on the grid?

@FT @doctorow Paraphrased: One venture capitalist who used to work at OpenAI told me that obsessing over who has the best model is no longer an issue... if OpenAI focuses on what it has, which is a consumer product with a huge audience, and can retain and monetize them, it will be successful

In @FT's briefing this morning, the inevitable logic of enshittification rears its head at OpenAI: pocketcasts.com/podcast/ft-new… h/t @doctorow obviously!

Sources: US households: EIA eia.gov/energyexplaine… Costco: Dave Larson linkedin.com/feed/update/ur… Gen capacity: NERC nerc.com/globalassets/o… Data centers: Cleanview distilled.earth/p/these-data-c…

- The top 9 data center projects by capacity (from Cleanview) add up to ~26 GW of demand. So -- if we were powering them with what's online today, those 9 projects would use 1/30th of the country's electricity generation. Whoa.

It can be hard to hold onto a sense of scale when we're talking about the grid. A few references:: - Each US household uses ~1 kW of demand. - One Costco uses ~1 MW of demand (not sure if that's average or peak!) - There's 1,000 GW of generation today, growing fast

More to come from us on this! NOVEC's 2025 large load request to PJM: pjm.com/-/media/DotCom… NERC's LTRA here: nerc.com/globalassets/o…

Thinking through data center connection in this way changes the implications of NERC's graph -- rather than telling us where resource adequacy risks are rising, it might tell us instead where data centers connections are more likely to be delayed.

I skimmed through the NERC Long-term Reliability Assessment that came out last week. Lots to dig into in there -- but I think we can all agree that grid conditions are getting tight amidst forecasts of a data-center-driven demand boom in the short term.