To operate reliably, the US electrical grid needs to balance supply and demand: to make sure, at any given moment, that the amount of electricity demanded by homes, businesses, and factories is equal to the amount being supplied by nuclear reactors, gas turbines, and other types of power plants.
Agreed. Enhanced geothermal is more speculative, but offers some of the same benefits as nuclear: cheap heat, spinning turbine, and baseload power with good location flexibility.
Nuclear is good (once you get it built) at providing a steady amount of base generation. It needs to operate at a high capacity factor, so it can't be ramped up and down to meet load. You need more flexible generation sources to swing to meet load, provide ancillary services, etc.
I need to read more about molten salt storage. A priori, storing something required to stay hot seems less efficient than pumped hydro or batteries, but again, I'm ignorant. On SMRs, I think utilities are reluctant to build until we see more come online and have a better idea of costs and timelines
Each connection has to be reviewed. How will it affect the current grid? What are the lines available and how much can they carry? How will the power be balanced against demand?
Beemac's answer about the process is good, but also consider the reason: an AC grid spanning 1000km or more is a fragile thing. Everything has to cooperate in balance and respond properly to disturbances. So while part of the process is NIMBY, there is also part of the process which is a necessary expert review.
This is excellent but omits a critical component of adding batteries to the grid. COST. Batteries are still very expensive. Have you modeled out what an "optimal" grid looks like with full battery deployment and how that impacts the cost per KWh? Thanks.
EDIT: the take home point is "At $400 a kw solar and $100 a kwh batteries (costs China is probably achieving right now), we could meet 80% of electricity demand with solar PV for roughly current US average combined cycle gas turbine costs"
Note that it doesn't include nuclear, wind, and hydro (which constitute ~35% of US electricity generation today but could grow) in the energy mix.
For a more comprehensive approach, report proposes a path to a "90% clean" electricity generation mix that's slightly cheaper than today's electricity:
Plus PV and battery costs per unit of power or energy are still dropping -- they're _not_ yet fully mature technologies where we can only eke out modest gains on cost or quality.
Texas likely spent 5.7 .. 7.2 $B to make the 4GW extra capacity mentioned in the post, which saved consumers $750M in the first year. That is better than a 10 year ROI which in utilities is amazingly good, and the costs are steeply falling every year. And that is for arbitrage: you only need about 15% of the battery for grid stabilization.
Batteries have caused some serious grid oscillations in the western US. Also, Chinese inverters have been found to have remote control capabilities so they can damage and destabilize the grid.
It does blow my mind that we're installing enormous numbers of internet-connected Huawei inverters. Insert Archer "do you want ants?" meme, except replace ants with collapsed grids.
I suspect that issue will get resolved with performance standards for the firmware that runs the inverters. Tesla already has put a lot of work into meeting standards in Australia and the EU for making battery inverters look to the grid like a spinning chunk of metal, providing "virtual inertia". A quick google turns up a couple articles from as long ago as 2022 on that work:
Performance has only improved since then. Because the frequency response logic is built directly into the embedded board operating the MOSFET gates, at like 40 kHz or higher, you can get frequency response behavior in less than a cycle.
What happens if we suffer another Carrington Event, or our enemies figure out how to send an EMP to the grid? We already have a transformer shortage, a rare Earths shortage too, with China the main supplier of these critical electrical materials. As we go all-electric in everything from the grid to cars to ovens and appliances, what is our protection against inevitable disruptions, natural or manmade?
We don’t have a rare earths shortage at all. The US has the largest proven reserves in the entire world, in the Great Basin, and we aren’t even mining them. All that’s happening is that demand is outstripping supply, which will do what it always does: bring more production into the market. But we’re in no danger of running out whatsoever.
The problem has been that China was willing to invest in processing the lanthanides and as it was a relatively small industry they put most of the competition out of business. You can get the ores from around the world but investing in refining ore and separating the elements has been a messy business and will take time to construct.
I think these events would damage the grid regardless of energy source. Long power lines and grid equipment are the point of failure, not energy generation per se.
More localized energy production via batteries and renewables probably makes the grid more resilient. There are ways to harden the grid to these events and build more transformers. We just don't because of bad policy choices.
Solar and lithium iron phosphate batteries don't use rare earths or any hard-to-obtain resources.
The transmission lines would be damaged but also have automatic disconnects. A Carrington or EMP creates huge potential over long distances like transmission lines but if they disconnect you will not necessarily get large damage to consumers or equipment at the ends of the lines which will pick up less. Cities with buried power should be good, rural areas with last-stages on poles might be vulnerable, but for those it should be a similar problem to lightning, which they do handle. Bringing it all back on line will take a while.
I'd like to see a simulation for a global HVDC grid connecting large solar farms around the world. My bet is the HVDC links are much cheaper than the comparable storage, while shifting the daily and seasonal solar cycle by linking places like California with the Gobi desert and Australia, to the point where almost no storage is needed.
For example, the new Australia - Singapore HVDC interconnector is 4300 km long and has a power rating of 2.4 GW for about $6 billion in cost. Assuming it can operate 8h a day for 20 years without significant refurbishment, that's about 4c/kWh. I doubt batteries can beat that even for daily discharge cycles. When you factor in the seasonal effects (basically, it's always sunny somewhere on Earth), the case for a world wide super grid becomes unbeatable, and the utilization of the cable increases further, since you will both export your power during the day while importing from other time zones during the evening/morning/night.
Incidentally, 4000km is about the distance from Spain to Nova Scotia, perhaps even New England. Since cable installation costs don't scale linearly with power, it should be possible to link USA to Europe with a 10-20GW link with something like a few tens of billions, which would pay itself back in a few years:
Nice work again ... it is curious to me that the "sunshine" state.. Florida is so far behind on Utility Solar/Battery. Florida is second for "roof-top" solar..
I wonder if the little spike in battery discharge just after midnight is meeting demand from a large number of EVs that are programmed to start charging at ultra-off-peak ToU prices at midnight?
Excellent work, thank you. The graphic labelled "CAISO Battery Storage Activity on June 1st 2025" (and also for '22) has an absurd X-axis, as you can see on CAISO's site: https://www.caiso.com/todays-outlook/supply#section-batteries-trend (and set the date widget to June 1). The shape looks right: maybe the time is UTC?
-How are batteries making the grid MORE reliable. It would seem they are able to displace existing more scalable technologies (peaker plants can run for a very long time)
-Given they are displacing scalable technologies, what happens if the peaker plants are economically forced off the grid. Will the grid still be reliable?
The NERC graphic you have for BESS response curve, compared to spinning generators, is properly labeled as the response of grid-following inverters. The batteries themselves are sub-second - and they can support every color of the rainbow in that image, out to a couple of hours, while the spinning inertia of generators runs out in 5 seconds.
So a redo of that chart with grid-forming inverters would show a nearly perfect flat line at 60 Hz.
Great intro! Would love to see you write on what is bringing battery costs down so much (big reason as to why they are growing of course) and on what the revenue and financing outlook looks like!
Unfortunately it is apples and oranges - grid connected batteries are not required to move. The only real problem with battery powered aeroplanes is the weight.
There are some promising avenues for low weight batteries which might allow electric only passenger planes, but this is decades away at best.
100%. The energy density issue compounds with distance. Batteries lose very little weight with discharge, while av gas burden on the plane decreases as it is consumed.
You do have to remember that you can't just count the weight/volume of the fuel, you have to count the weight/volume of the system that converts the fuel into thrust, as well. It turns out that electric motors win over engines and turbines, on that side of things, though not by enough to swing the overall calculation in their favor yet.
David Roberts has talked with people who are trying to attack that problem.
Batteries are a promising fix to electrical sustainability. I have a question about generation.
If the problem is gas/coal/solar/wind generation, why isn't Nuclear being thrown in the mix?
Nuclear provides more power, cheaper, and it's easy to maintain, especially the new salt and boron reactors.
Matt Yglesias has written a number of columns about how dysfunctional regulation has mired development of commercially-viable new reactor models.
https://www.slowboring.com/p/the-nuclear-policy-america-needs
https://www.slowboring.com/p/can-america-get-to-yes-on-a-new-reactor
https://www.slowboring.com/p/unleash-the-power-of-bipartisan-nrc
Agreed. Enhanced geothermal is more speculative, but offers some of the same benefits as nuclear: cheap heat, spinning turbine, and baseload power with good location flexibility.
Nuclear is good (once you get it built) at providing a steady amount of base generation. It needs to operate at a high capacity factor, so it can't be ramped up and down to meet load. You need more flexible generation sources to swing to meet load, provide ancillary services, etc.
Some of the SMRs in development address this issue by incorporating molten salt heat storage.
I need to read more about molten salt storage. A priori, storing something required to stay hot seems less efficient than pumped hydro or batteries, but again, I'm ignorant. On SMRs, I think utilities are reluctant to build until we see more come online and have a better idea of costs and timelines
Why is the interconnection queue so long?
Each connection has to be reviewed. How will it affect the current grid? What are the lines available and how much can they carry? How will the power be balanced against demand?
Beemac's answer about the process is good, but also consider the reason: an AC grid spanning 1000km or more is a fragile thing. Everything has to cooperate in balance and respond properly to disturbances. So while part of the process is NIMBY, there is also part of the process which is a necessary expert review.
This is excellent but omits a critical component of adding batteries to the grid. COST. Batteries are still very expensive. Have you modeled out what an "optimal" grid looks like with full battery deployment and how that impacts the cost per KWh? Thanks.
You may be interested in this construction physics post on supplying the grid with solar, batteries, and natural gas.
https://www.construction-physics.com/p/can-we-afford-large-scale-solar-pv
EDIT: the take home point is "At $400 a kw solar and $100 a kwh batteries (costs China is probably achieving right now), we could meet 80% of electricity demand with solar PV for roughly current US average combined cycle gas turbine costs"
Note that it doesn't include nuclear, wind, and hydro (which constitute ~35% of US electricity generation today but could grow) in the energy mix.
For a more comprehensive approach, report proposes a path to a "90% clean" electricity generation mix that's slightly cheaper than today's electricity:
https://www.2035report.com/electricity/
Plus PV and battery costs per unit of power or energy are still dropping -- they're _not_ yet fully mature technologies where we can only eke out modest gains on cost or quality.
Texas likely spent 5.7 .. 7.2 $B to make the 4GW extra capacity mentioned in the post, which saved consumers $750M in the first year. That is better than a 10 year ROI which in utilities is amazingly good, and the costs are steeply falling every year. And that is for arbitrage: you only need about 15% of the battery for grid stabilization.
Batteries have caused some serious grid oscillations in the western US. Also, Chinese inverters have been found to have remote control capabilities so they can damage and destabilize the grid.
It does blow my mind that we're installing enormous numbers of internet-connected Huawei inverters. Insert Archer "do you want ants?" meme, except replace ants with collapsed grids.
Wait till China decides to cause roof top solar panels to overload and burn. There goes the neighborhood.
How on earth would a battery cause grid oscillations?
Do you have a link?
I will get the link to the report. It’s the inverters that caused the oscillations. Batteries are DC and the grid is AC.
I suspect that issue will get resolved with performance standards for the firmware that runs the inverters. Tesla already has put a lot of work into meeting standards in Australia and the EU for making battery inverters look to the grid like a spinning chunk of metal, providing "virtual inertia". A quick google turns up a couple articles from as long ago as 2022 on that work:
https://electrek.co/2022/07/27/tesla-virtual-machine-mode-coming-big-battery/
https://cleantechnica.com/2022/08/01/world-1st-tesla-batteries-providing-inertia-services-at-scale/
Performance has only improved since then. Because the frequency response logic is built directly into the embedded board operating the MOSFET gates, at like 40 kHz or higher, you can get frequency response behavior in less than a cycle.
Link: https://www.naspi.org/sites/default/files/2022-04/20220330_naspi_webinar_QA_answers.pdf
What happens if we suffer another Carrington Event, or our enemies figure out how to send an EMP to the grid? We already have a transformer shortage, a rare Earths shortage too, with China the main supplier of these critical electrical materials. As we go all-electric in everything from the grid to cars to ovens and appliances, what is our protection against inevitable disruptions, natural or manmade?
We don’t have a rare earths shortage at all. The US has the largest proven reserves in the entire world, in the Great Basin, and we aren’t even mining them. All that’s happening is that demand is outstripping supply, which will do what it always does: bring more production into the market. But we’re in no danger of running out whatsoever.
The problem has been that China was willing to invest in processing the lanthanides and as it was a relatively small industry they put most of the competition out of business. You can get the ores from around the world but investing in refining ore and separating the elements has been a messy business and will take time to construct.
I think these events would damage the grid regardless of energy source. Long power lines and grid equipment are the point of failure, not energy generation per se.
More localized energy production via batteries and renewables probably makes the grid more resilient. There are ways to harden the grid to these events and build more transformers. We just don't because of bad policy choices.
Solar and lithium iron phosphate batteries don't use rare earths or any hard-to-obtain resources.
The transmission lines would be damaged but also have automatic disconnects. A Carrington or EMP creates huge potential over long distances like transmission lines but if they disconnect you will not necessarily get large damage to consumers or equipment at the ends of the lines which will pick up less. Cities with buried power should be good, rural areas with last-stages on poles might be vulnerable, but for those it should be a similar problem to lightning, which they do handle. Bringing it all back on line will take a while.
I'd like to see a simulation for a global HVDC grid connecting large solar farms around the world. My bet is the HVDC links are much cheaper than the comparable storage, while shifting the daily and seasonal solar cycle by linking places like California with the Gobi desert and Australia, to the point where almost no storage is needed.
For example, the new Australia - Singapore HVDC interconnector is 4300 km long and has a power rating of 2.4 GW for about $6 billion in cost. Assuming it can operate 8h a day for 20 years without significant refurbishment, that's about 4c/kWh. I doubt batteries can beat that even for daily discharge cycles. When you factor in the seasonal effects (basically, it's always sunny somewhere on Earth), the case for a world wide super grid becomes unbeatable, and the utilization of the cable increases further, since you will both export your power during the day while importing from other time zones during the evening/morning/night.
Incidentally, 4000km is about the distance from Spain to Nova Scotia, perhaps even New England. Since cable installation costs don't scale linearly with power, it should be possible to link USA to Europe with a 10-20GW link with something like a few tens of billions, which would pay itself back in a few years:
https://cleantechnica.com/2024/12/16/hvdc-transmission-between-europe-north-america-makes-fiscal-sense/
What about sabotage risks?
I imagine they are comparable or lower than gas pipelines - since that technology is less robust physically and less energy dense.
So, pretty well understood; but significant, especially if you can disrupt a world spanning coalition of countries that oppose your goals.
Nice work again ... it is curious to me that the "sunshine" state.. Florida is so far behind on Utility Solar/Battery. Florida is second for "roof-top" solar..
I wonder if the little spike in battery discharge just after midnight is meeting demand from a large number of EVs that are programmed to start charging at ultra-off-peak ToU prices at midnight?
Excellent work, thank you. The graphic labelled "CAISO Battery Storage Activity on June 1st 2025" (and also for '22) has an absurd X-axis, as you can see on CAISO's site: https://www.caiso.com/todays-outlook/supply#section-batteries-trend (and set the date widget to June 1). The shape looks right: maybe the time is UTC?
Ah, it's rendering correctly on the webpage.
There seems to be two obvious questions.
-How are batteries making the grid MORE reliable. It would seem they are able to displace existing more scalable technologies (peaker plants can run for a very long time)
-Given they are displacing scalable technologies, what happens if the peaker plants are economically forced off the grid. Will the grid still be reliable?
Great article. Very informative, Thanks 👍
The NERC graphic you have for BESS response curve, compared to spinning generators, is properly labeled as the response of grid-following inverters. The batteries themselves are sub-second - and they can support every color of the rainbow in that image, out to a couple of hours, while the spinning inertia of generators runs out in 5 seconds.
So a redo of that chart with grid-forming inverters would show a nearly perfect flat line at 60 Hz.
Great intro! Would love to see you write on what is bringing battery costs down so much (big reason as to why they are growing of course) and on what the revenue and financing outlook looks like!
Maybe am comparing “apples to oranges” if batteries are effective for electric grids why are they not effective for airplanes?
Unfortunately it is apples and oranges - grid connected batteries are not required to move. The only real problem with battery powered aeroplanes is the weight.
There are some promising avenues for low weight batteries which might allow electric only passenger planes, but this is decades away at best.
Excessive weight, probably. The energy density of oil-based fluids (such as kerosene) is hard to beat.
100%. The energy density issue compounds with distance. Batteries lose very little weight with discharge, while av gas burden on the plane decreases as it is consumed.
You do have to remember that you can't just count the weight/volume of the fuel, you have to count the weight/volume of the system that converts the fuel into thrust, as well. It turns out that electric motors win over engines and turbines, on that side of things, though not by enough to swing the overall calculation in their favor yet.
David Roberts has talked with people who are trying to attack that problem.
https://www.volts.wtf/p/whats-up-with-electric-aviation
https://www.volts.wtf/p/whats-up-with-hydrogen-electric-aviation
Looking at back of the envelope math, this seems like a place where hydrogen is actually pretty promising.