And pack-level Li ion prices have now dropped to almost half what the cell-level prices were in 2018 on your graph.
AI data centers are forcing dirty ‘peaker’ power plants back into service
- Thread starter Daniel Nenni
- Start date
And pack-level Li ion prices have now dropped to almost half what the cell-level prices were in 2018 on your graph.
Sorry, I forgot to mention the huge caveat to my numbers was going DIY, outsourcing the work to a solar contractor totally obliterates the numbers in the US; we have the worst overhead numbers for residential solar deployment in the world by far. Typical 10kw system quote in the US from a solar contractor is $28k-35k, same system in a developed country like, say, South Korea goes for $4k, labor cost inclusive. What's obscene is the customer acquisition cost in the United States is nearly a dollar a watt...$10k of that $28k-$35k figure simply goes into marketing costs to acquire the customer. 10kw system produces more energy than the average American home uses for just about anywhere in the lower 48 states for context. We went all electric in our household, payback was just 2 years, and been saving us around $10k a year, though since we're off-grid, a grid connection wasn't even remotely an option for us to begin with. But building out the system made us completely worry free about energy, I can leave everything on without worrying about a power bill, it definitely changes our living habits, for better or worse!Not for me. My electric bill is about $2K/year, estimating high. I likely won't live long enough to get to breakeven. (My home is just under 4K sqft.) I also spend about another $1.5K on natural gas for heat and hot water, but going to all electric would cost about $30K. No way that pays off.
If we had two electric vehicles, that would move the pure energy breakeven point earlier, but I'm not a fan of current EV technology, and two EVs for us would be at least $100K. That isn't happening either.
Yeah, but there's a lot more to solar generation than just panels and batteries. You need about 5000 acres of land per gigawatt, correct? In a lot of highly populated areas of the US (like the northeast) you don't see a lot of sun in the winter either.
I am anxious to see the SMR numbers, if I live long enough.
Great analysis, but not entirely convincing.
I agree with you on the EV part for my specific needs, but then again, it's still the very early stages of EV development and adoption if we compare with ICE vehicles, I'm fairly confident that most of the shortcomings will be addressed within the next 15 years (i.e. recharge speeds, cold weather operation, endurance, energy density, cost). I can't get a reasonably priced EV truck with enough range to haul livestock feed up to my place, but 5...10 years from now? Decent chance we'll get there, all the more if I consider used options. But even for the vast majority of Americans today who just drive from home to work every day, a modestly priced EV isn't out of reach, can provide more than enough range for a daily commute, and the beauty of electricity is, it is something you can produce at home unlike liquid fuels. Heck, I was looking to DIY a small scale Fischer-Tropsch reactor to produce liquid fuels at my home just to take advantage of all the excess solar power I have in the summer to use to heat my home in the winter but unfortunately it costs more than just...getting a bigger battery
.There certainly is more than just panels and batteries, and as you rightfully point out, land use is the primary downside though at least on that end, the US has a lot relatively to many other small industrial nations. I certainly wouldn't suggest going RE100 for countries like Taiwan or Korea, that's just insanity, but unfortunately the governments of both aforementioned countries seem to be making a huge push towards that because of all the international pressure. 5000 acres for a gigawatt sounds substantial, but say, less than 20% of federal BLM land in Nevada would be sufficient to produce all of the energy needs of the entire US. Not that I'm suggesting that due to transmission losses, but just providing context...and by the way, that 10kw setup for an average home would only require a little over 400sqft of roof space. I've consulted thousands of households for DIY solar and the New England states were great for solar in terms of ROI; electricity is so expensive out there, it's cost effective to just double the size of the solar array to factor in lower generation in the winter. The only place where ROI was challenging was western Washington, low solar irradiation combined with cheap electricity from the Columbia River Basin was the only place where DIY solar + batteries ROI extended past 10 years, but pretty much everywhere else, it's a bit shorter than that. Solar panels are so cheap, I'm building a carport right now with solar panels for the roof and siding, buying used panels are cheaper than plywood, sheet metal, vinyl or any other siding material while being more durable, rain/snow proof, freeze proof not to mention it can produce electricity. 10 cents per watt for panels that still have plenty of life left so when EV trucks do get cheap enough, I won't have to pay for fuel!
As for SMRs, all the projections show SMR's overnight construction costs being more costly than conventional nuclear power plants in the near future. The three major contenders for SMRs in the US, NuScale, TerraPower and X-energy, they've all partnered with Doosan Enerbility for reactor vessels, pressure vessels and components, I trust South Korean companies can execute and produce everything in a timely, cost effective manner, but putting it all together in the US is going to be...entertaining. NuScale's plant in Idaho was cancelled when projected costs soared from $3.6b to $9.3b for a 462MW 6 reactor facility, or over $20/w. TerraPower's Natrium plant is expected to be at least $4b for a 345MW plant, or a little over $11.50/w, but I'll believe it after it's all said and done (there's a quote out there estimating cost for the reactor in Kemmerer to be $10b, or $29/w). Xe-100 from X-energy, no numbers published, actual or projected other than claims of under $60/MWh-e LCoE. Meanwhile, in 2024, South Korea budgeted under $9b for two APR-1400 reactors at Shin-Hanul, or $3.2/w. GE-Hitachi's BWRX-300, $18/w according to TVA's filings. Having followed every nuclear power plant opened in the past decade and planned for the next around the world, I have absolutely no confidence in nuclear power plants, SMR or conventional, being built for under $10/w in the west. I am hopeful the drive for AI will at least inject some much needed capital for R&D in this sector and allow some reactors to be built, but as with many other technologies originally developed in the west, we'll fail to commercialize it and countries like China and Korea will take what was learned from west and run with it. Hate to be a pessimist when I am so pro-nuclear, hopefully I'll be proven wrong?
Yep, I only used that graph because it showed prices for the past few decades, but actual pack figures are even lower than your chart for utility scale ESS applications since it includes vehicles. And keep in mind a lot of projects in the US still use NCM chemistry, utility scale LFP ESS in China right now average $60/kwh, all costs inclusive for whole system. Kind of like solar, the battery cells themselves have become cheaper than the rest of the ESS, solar modules are $0.07/w for a container, FOB China for any tier 1 supplier, good quality inverters cost more than that per watt, even mounting rails, wiring/conduit cost more than that per watt, they're no longer the priciest component in the system. With batteries, I'm getting tier 1 suppliers quoting around $30/kwh for even modest quantities, the rest of the BOM (i.e. housing assembly, busbars/wires, BMS, etc.) cost more than the raw cells.
CATL announced $19/kwh for their sodium ion cells, they're worse than LFP in almost every metric now with the exception of cold weather performance, but if they're quoting that now and I'm hearing projections of $10/kwh by the end of the decade...that means $30/kwh all inclusive utility scale ESS not too far away at which point, even $10/w nuclear power plants can't compete in all but the smallest geographies. And frankly, I'm not much of a proponent of hydrogen, but once you have solar get so cheap, you can brute force your way through conversion inefficiency for hydrogen or other liquid fuels (ethanol, methanol, ammonia, synthetic gas/diesel) a la Fischer-Tropsch, Sabatier or Haber processes. Batteries storage full and tons of solar idling? Just use that excess capacity to produce fuel. We're not going to have intercontinental airliners or container ships powered by batteries in our lifetime, but that could be a viable path to decarbonizing those sectors. There's 1.5TW of solar module production capacity as of this year in China alone. US uses about 500GW of electricity 24/7. US capacity factor for solar ranges from 13.5% in December to 31% in July. 4TW of solar would provide all the electricity used in this country to cover the shortest months. That's under $300b in module costs and just over 2.5 years of China's module production capacity. It's no longer a rounding error and that's saying a lot from a guy who at one point worked in oil & gas!
We're spending $123m on a 122 unit homeless housing project in Santa Monica, over $5 billion on a 2.25 mile people mover at LAX (almost half a million dollars per foot!), and as much as I like high speed rail, as of last year, $106.2b projected cost for the first phase of California's HSR...which based on current progress and cost increases, I don't see it being completed by 2035 for a buck less than $250b. My concern with these datacenters, they'll build them where electricity is cheapest. Expect rates to go up across the board for all Americans soon, with those who pay the least per kwh to be impacted the most. Fortunately for Californians who live on the grid operated by an IOU, probably won't impact them much because they're already getting screwed over so much!
