Astrid Atkinson

Flexible Data Centers: Research Report Research by Camus, encoord, & Princeton ZERO Lab shows that flexibility unlocks faster data center interconnections and helps internalize 100% of system costs.

Camus has a new white paper out today! Following on from earlier work on the potential to leverage existing grid capacity to connect large loads faster & more cheaply, we took the next step to model load flexibility at a set of real sites with real utility data.

www.camus.energy/flexible-dat...

Also, huge shout-out to our partners Encoord and Princeton ZERO lab, and our financial backer Google. And to my own team members, who worked hard to get this out into the world. Couldn't ask for a better team.

I would be remiss if I did not add that Camus can help you out with this. Our platform makes it easy to model & operate flexibility at one site or many sites. We want to get datacenter sites flexibly connected to the grid, helping them get the power they need and reducing cost and carbon for all.

Right now, the datacenter power conversation is all over the place - either focused on full flexibility (which most sites can't offer, though props to those which can) or fully off-grid new fossil generation. Neither option is quick. Site level flex which allows for a mix gets everyone unstuck.

This allows us to get a lot more utilization out of the grid we've already built - both capacity, and poles and wires. They can handle it most of the time! Our modeling shows less than 1% flexibility required in our test sites to enable them to fit the existing grid.

This model does require some kind of market structure to support contingent flexible load. In our modeling we assumed a mix of firm and flex capacity. This gives the site a floor for how much local flexibility they might need to provide, but allows them to use excess grid capacity most of the time.

This also matters for affordability. The market monitor cited huge capacity cost increases, passed on to consumers, for adding datacenter loads. If you allow the datacenter to bring some of their own capacity through onsite resources (effectively reducing site load), those costs impacts vanish.

But load curtailment is just one option in the flexibility stack (and some datacenters do want to offer it). Onsite generation and storage make sense for most datacenters, and most also want a grid connection. Bringing an "all of the above" approach actually requires flexible interconnection.

The recent PJM market monitor report made waves calling datacenter flexibility a "regulatory fiction". That's true in the narrow sense that there are not many sites lining up to turn off their servers in an unbounded way (PJM stakeholders just rejected a fully conditional large load model).

This works because most locations have *some* grid power available, but few locations have enough existing headroom to connect an entire new large site without any major upgrades. Flexible connections enable datacenters to bring some of their own onsite power and get some from the grid.

The biggest takeaway: site level flexibility can speed up time to power (for both generation and wires capacity) by years. Sites which require transmission upgrades or new generation to be built could cut timelines from 5-7 years to 2-3.

Not included in the final paper, but also very interesting to our team: what's the carbon optimal stack? For most of the sites studied, the additional cost of providing flexibility entirely through onsite battery and avoiding new fossil gen was not much higher (hoping to do more work on this soon.)

In our paper, we modeled the specific flexibility required across a range of sites, and the cost optimal flexibility stack options (from load reduction to onsite generation) which could meet that need.

We also took a broad view of what load flexibility means - people think of reducing load within the datacenter, which is one potential mechanism, but any datacenter that wants to bring onsite generation or battery as part of their power mix also fits this model.

We see this work as filling in an important piece of the puzzle - flexibility offers great promise, but what would that look like in practice? When does a site curtail, for how long, how much, and how often?

Flexible Data Centers: Research Report Research by Camus, encoord, & Princeton ZERO Lab shows that flexibility unlocks faster data center interconnections and helps internalize 100% of system costs.

Camus has a new white paper out today! Following on from earlier work on the potential to leverage existing grid capacity to connect large loads faster & more cheaply, we took the next step to model load flexibility at a set of real sites with real utility data.

www.camus.energy/flexible-dat...

Totally agree, few things better than just telling someone that what they have done is amazing.

Slightly left field recommendation, "The Jazz of Physics" by Stephon Alexander. It's a bit of a memoir, a lot about jazz, and mostly about the origins of the universe, written by a quantum cosmologist who specializes in the physics of creation/origination.

Wow, an extraordinary generous offer from a Bolts reader:

Russ is offering to match any contribution you make to Bolts (just let him know)!

And he'll also throw in an additional $$ for each repost of his post, or each new follow @boltsmag.org gets. So at least smash those share or follow buttons!

Done 😀

As a lifelong reader, romance is the only genre I still read compulsively (like, 2-3 novels a week). There's a lot to be said in this stressful world for books where people are definitely going to be happy in the end.

I see the renewables industry focusing on permitting reform and "all of the above" energy, confident that renewables will win on price in a level playing field. And then I see fossil interests tilting the board.

Rogue One is my favorite movie of all time, thank you for your attention to this matter.

If it doesn't help right away (like, you walk out of the first visit feeling a lot better), don't throw good money after bad.

FWIW I have a really good PT / sports chiro who does myofascial work and she fixed my persistent lower back pain in about 4 visits.

Yes.

All I want: give me the option to pay for the equivalent of one magazine or newspaper, in order to read one story. I'd pay $5. I'm not going to buy a monthly subscription.

She knows enough to knock your hands off them though!

Stop doomscrolling and look at these capybaras.

I have a marble veneer fireplace. Someday it'll get replaced, it's ugly as hell.