BloombergNEF Blog

> Is it? I don't think so. I mean, they're projecting $750 billion in 2026, and apparently they spent $450 billion on them last year: https://about.bnef.com/insights/commodities/ai-data-center-build-advan… $25 billion is like six datacenters worth of money: https://www.reuters.com/commentary/breakingviews/how-big-techs-630-bln… It's a drop in the bucket.

The absolute best thing I've read on this subject is this article here: https://about.bnef.com/blog/liebreich-generative-ai-the-power-and-the-… It talks at great length about data center trends relating to generative AI, from the perspective of someone who has been deeply involved in researching power usage and sustainability for two decades. I made my own notes on that piece here (for if you don't have a half hour to spend reading the original): https://simonwillison.net/2025/Jan/12/generative-ai-the-power-and-the-…

Short term rise due to commodity prices, but shifts to LFP and other cheaper approaches will allow the dramatic long term price trend to continue. https://about.bnef.com/blog/battery-pack-prices-fall-to-an-average-of-… If you look at PV there were similar "ah ha, costs have stopped falling!" moments in the past (mid 2000s, for example), but they didn't last long and were followed by even more dramatic cost declines. It's important not to be selective here, like climate change deniers are looking at global temperature data.

Even the most hydrogen positive environmentalists don't think hydrogen makes sense for cars. Those pushing h2 cars either have natural gas interests, or an extreme attachment to fueling. By far the best current assessment of how we will use hydrogen as energy transport I have found: https://about.bnef.com/blog/liebreich-separating-hype-from-hydrogen-pa…

We already have that in the form of LiFePO4 batteries. They last over 2000+ cycles, use no rare/expensive/conflict minerals and are safe. Also just recently they crossed the $100/kWh barrier in some markets: https://about.bnef.com/blog/battery-pack-prices-cited-below-100-kwh-fo… An ISO container(40ft) of those can store 10-15MWh. For the cost of the Astravets Nuclear Power Plant in Belarus ($11bln, 2GW) you could have 80GWh worth of batteries - that's a week of backup power for one million households in the EU. One disadvantage of them has always been their performance in terms of power and energy density, but given that they're currently installed in Chinese-made Teslas, I guess it's more or less a solved problem.

great read on this topic - https://about.bnef.com/blog/behind-scenes-take-lithium-ion-battery-pri… As usual, more nuanced than I naively thought. "Put simply, something that looks like a future breakthrough today might end up simply being another point on the curve, by the time it is developed and engineered into a viable product and brought to market in a decade’s time."

It irks me to see predictions talked about as if past events. But this is a more solid guess than most. Actual progress has been impressive: 85% reduction in the 2010-18 period, says the article. Graph here: [1] Getting another 50% in 10 years can probably be done from manufacturing scale alone. Giant battery pack factories seem to be going up all over. This is something that ought to be almost totally automated, but doesn't yet seem to be. You're making large numbers of identical units for years per model, the best case for automation. Somebody will get that right soon. Do we get more energy density? 2x more, and gasoline cars are toast. [1] https://about.bnef.com/blog/behind-scenes-take-lithium-ion-battery-pri…

The path to vehicle electrification might not be that long: https://about.bnef.com/electric-vehicle-outlook/ Also, it's worth making the distinction between vehicles and vehicle miles travelled. We might not replace all the cars right away, but the majority of the miles travelled might convert pretty quickly if they're done via TCO sensitive fleet vehicles (Uber, Lyft, etc) and/or 2nd car commuter vehicles.

"Cumulative passenger EV sales are set to hit 4 million this week. This includes battery electric passenger vehicles and plug-in hybrid passenger vehicles. If we include e-buses, the 4 million milestone was reached at the beginning of July. The time needed to reach each consecutive million EVs sold has shrunk from 17 months for the second million to 6 months for the fourth million. The next million EVs will take just over 6 months. We expect the five-millionth EV to be sold in March 2019." I won't quote the entire article here, but it's worth a few minutes of time. The EV sales rate is climbing rapidly; not exponentially, but much faster than linearly. https://about.bnef.com/blog/cumulative-global-ev-sales-hit-4-million/

If that's not a scam or something that fails in 2 years, it's pretty amazing. At a lifetime of 3500 full-depth cycles it works out to a lifetime capital cost of only 2.2 cents per kWh. In sunny regions, with instantaneous solar generating costs already below 2.5 cents/kWh, it means that you can supply power around the clock at costs lower than fossil combustion. In a March 2018 report, BNEF put 2017 utility-scale battery costs at $209/kWh: https://about.bnef.com/blog/tumbling-costs-wind-solar-batteries-squeez… And that report was already taking into account global supply chains. That's why I am skeptical about the low Alibaba price. If EPCs can get reliable battery packs at $78/kWh (sans inverter) just by ordering from China, I would expect utility scale battery installation rates to soon go from "brisk" to "breakneck."

no place has as high a capacity factor for solar pv as 50%; the world average is 14% (i.e., 14 watts average output per 100 watts nameplate output) and the country with the highest is egypt at 35% (this doesn't mean that you get zero power for 65% or 86% of the day. it means that you get lower than max power at all times except noon. to a great extent you can compensate by that by just installing more panels, but at night you need a better strategy) but the capacity factor is irrelevant to the fact that solar energy on the power grid generally sells for about half the price that coal-generated electricity sells for, on the same grid, when storage capacity is insufficient. in fact, prices used to go negative at night (to avoid shutting down slow-ramping baseload plants) and now they go negative in the day https://pv-magazine-usa.com/2020/05/28/record-low-solar-ppas-in-the-so… is an article from four years ago giving some specific prices: a solar ppa had just been signed for 15 dollars per megawatt hour, while the cost of production with coal at the san juan generating station was 44.90 dollars per megawatt hour, even though it was built right on top of a coal mine to save on shipping costs. that's why the san juan generating station has been decommissioned. if you look, you'll find stories like this all over the place, and solar panels now cost half of what they did when that story was written now, it's…

The battery market is a pretty interesting market to follow. Bloomberg New Energy Fund (NEF) published an interesting report a few months ago titled "China Already Makes as Many Batteries as the Entire World Wants". https://about.bnef.com/blog/china-already-makes-as-many-batteries-as-t… I like their reports because it injects facts into what is otherwise a very opinion driven public debate. And the facts are quite impressive. Some key points: - they are tracking about 6 twh of new battery production related investment for 2025 world wide - that's about 4x the projected demand of 1.5 twh for 2025; which is up from slightly under 1 twh last year (2023). So, they are signalling massive over production of batteries is about to happen. That's quite a change from the shortages companies were dealing with in the last few years. - they expect that to have some obvious effects on prices and some of those investments: not all these projects will make it and battery over supply will drop prices and endanger producers that can't do that. I would add to that that it is likely that wider availability of cheaper batteries is going to create new use cases as well. So their demand projection might be off. If prices drop low enough, a lot of household equipment might start coming with batteries, for example. Things like fridges and ovens for example. That's a market that barely exists now. But if prices drop enough, why not? Another thing I find interesting is the…

Yeah but you can do the calculations yourself. I'm gonna use the most expensive storage technology I know (lithium-ion) and forego cheaper alternatives like pumped hydro just to make a point. I'm gonna use data from lazard, because it looks like the most recent study[1]. - Cost of Lithium-Ion Battery Packs 139$/kWh [2] - 139000$/MWh - We want the levelized cost of electricity, so let's assume, again to make a point, that we have to throw the battery out after 500 cycles instead of it only operating at 80% charge. That's 139000$ / 500 = 278$ - The EU estimates that in a 80% renewable grid you need approx 30% storage capacity. So let's assume we need 40% storage capacity for a 100% grid.[3] That's 278$ * 0.4 = 112$ - Cost for solar 24–96$ - Cost for wind 24–75$ - Cost for solar + storage 136$ - 208$ - Cost for wind + storage 136$ - 187$ - Cost for new nuclear: 141–221$ So you can combine renewables with the most expensive/deluxe storage technology available today and it's still cheaper than nuclear. 1: https://www.lazard.com/media/typdgxmm/lazards-lcoeplus-april-2023.pdf 2: https://about.bnef.com/blog/lithium-ion-battery-pack-prices-hit-record… 3: https://energy.ec.europa.eu/topics/research-and-technology/energy-stor…

200 billion PHP is 3.58 billion USD. The farm is 3500 MW and has 4000 MWh of battery storage. As of today 1 Wp (peak watt) of mainstream solar panels costs about 0.16 USD [1], and 1 kWh of li-ion battery costs about 140 USD [2]. Which means the panels alone cost 560M USD, and the batteries also cost 560M USD, so combined they represent 31% of the total 3.58B USD budget. It seems to be not very cost-efficient, as the overhead for everything else (2.46B USD) is quite large: inverters, mounting hardware, grid connection, installation, inspection, etc. Edit: I was basing my "not very cost-efficient" remark on the fact that this solar plant is basically $1.00/W however the best utility-scale solar plants achieve about $0.80/W, or 20% less. However I forgot to account for the fact this plant has batteries, so a higher overhead is expected. So, after all, the capex is in line with what I'd expect. [1] https://www.pvxchange.com/Price-Index [2] https://about.bnef.com/blog/lithium-ion-battery-pack-prices-hit-record…

This is incredible news! EVs provide the sweetest kind of demand for a grid, they are parked 23 hours a day, can charge whenever power is free and less than free. Power prices go negative 200 million times/year[1]. Batteries, Solar and Wind and still in the early stages, as volume grows, both will get a lot cheaper. Batteries add the much needed gigantic-distributed-storage-reservoir, soaking up all the excess production that is currently wasted. An insightful comment from a HNer (don't remember who), if there is excess production we have to either move it across time or across space. Across space requires building transmission lines (costly, takes a decade). Across time is easy, we just need more EVs, everywhere. https://en.m.wikipedia.org/wiki/File:3-Learning-curves-for-electricity… (2019 article) “Batteries will fall much faster than you are forecasting.” The key determinant of our forecast is the relationship between price and volume. From the observed historical values, we calculate a learning rate of around 18%. This means that for every doubling of cumulative volume, we observe an 18% reduction in price. Based on this observation, and our battery demand forecast, we expect the price of an average battery pack to be around $94/kWh by 2024 and $62/kWh by 2030: https://about.bnef.com/blog/behind-scenes-take-lithium-ion-battery-pri… Learning rate for Solar is 20%. According to this article (https://ourworldindata.org/learning-curve), the price declined…

Small reactors were part of the original 1st gen nuclear reactors. The world went to bigger reactors in the 60s-70s because it ended up being much cheaper than building lots of small ones. Reactor fabrication is complex, and it's easier to just build them bigger. From the 90s to the mid-2000s there was probably a window for development in this area. But the nuclear sector stagnated as a whole, and then fracking made natural gas cheap enough that people wouldn't consider it. The jury is out on SMRs still. They're promising a lot, and the tech and engineering has improved. But plunging costs of renewables mean the window of opportunity for nuclear tech is probably mostly closed in the West. This article from a highly respected energy sector analysis group is worth a read: https://about.bnef.com/blog/scale-up-of-solar-and-wind-puts-existing-c… India and China are the big markets for nuclear right now -- their energy demands continue to grow, and they're building reactors as part of an all-of-the-above model. France MIGHT be a potential buyer of SMRs (their reactor fleet is getting close to end-of-life), but their nuclear companies have put a lot of push behind the EPR design.

Batteries cost > $100 / kWh, the industry still considers that a target to brag about: https://about.bnef.com/insights/clean-transport/lithium-ion-battery-pa… This is WHOLESALE BOM price, which is about as useful as the price of uranium. That is horrifically expensive.

$100 / kWh capacity, over 10000 cycles, is $0.01 kWh delivered. It's dirt cheap. Also that's not the cost though. You're looking at vehicle packs not stationary storage packs at that cost. We've seen stationary storage packs get down to $50/kWh last year - with an average pack price of $70: https://about.bnef.com/insights/clean-transport/new-record-lows-for-ba… And prices continue to decay. Those are the most expensive prices you'll ever see going forwards. And there's other benefits. Batteries distribute well, letting us lower grid related costs by making more constant use of the chokepoints in the grid for instance.

I think the key is to combine it with a strong, digitalized grid and a lot of BESS—a technology which is now getting progressively cheaper, just like PV. https://about.bnef.com/insights/clean-transport/new-record-lows-for-ba… I believe it is realistic to expect that, in combination with other renewable energy sources such as wind (which, for example, generates more energy at night than during the day), biomass, and hydropower—along with the high level of grid integration currently taking place in Europe—the share of renewable energy could reach 100 percent in 10 or 15 years. Provided there is the political will to do so.

As much as 23.1GW of capacity is under construction globally, split across 831 sites. The Americas region hosts 17GW of this capacity, across 311 locations. Europe, Middle East and Africa or EMEA as well as the Asia Pacific (APAC) region trail significantly. APAC has marginally more ongoing construction, with 3.2GW across 283 sites versus EMEA’s 2.9GW across 258. The US alone hosts 15.9GW of ongoing construction. https://about.bnef.com/insights/commodities/ai-data-center-build-advan…

https://theicct.org/publication/electric-cars-life-cycle-analysis-emis… https://www.sciencedirect.com/science/article/abs/pii/S095965262300269… (paywall) https://about.bnef.com/insights/clean-transport/no-doubt-about-it-evs-… is a writeup of https://www.bnef.com/insights/33557 which is paywalled.

BloombergNEF has over the years proven to have pretty solid forecasts. The current one about NEVs [1] has a few interesting points. Adoption of EVs is slowing down in the US due to policy changes but going to explode in countries like Vietnam because they are cheeper to buy an run. It is not BMWs and Mercs but Chinese brands. In Europe and the US the Chinese EVs are kept outside with the help of tariffs but that is just closing the eyes to avoid facing the inevitability. Battery technology, production and raw materials is all China. Last not least Europe is driving up KWh costs by an ideologically driven push for renewables which also doesn't help. [1] https://about.bnef.com/insights/clean-transport/electric-vehicle-outlo…

The Pixie dust is called China. BNEF is tracking 7.9 TWh of annual battery manufacturing capacity for the end of 2025 [1]. Chinese manufacturers' all-in costs for BESS are now down to $66/kWh and still dropping [2]. We (or at least China) have crossed the "knee" of the exponential for battery production, and loads of people don't seem to realize this. [1] https://about.bnef.com/blog/china-already-makes-as-many-batteries-as-t… [2] https://cleantechnica.com/2024/12/24/what-are-the-implications-of-66-k…

The world invested $2.1 trillion in the clean energy transition in 2024. We are doing the big things. https://about.bnef.com/blog/global-investment-in-the-energy-transition…

This is up front price, not TCO where it's already parity. Also the 50% number is based on a bump in cathode material price in 2022 and 23 as can be clearly seen in the graph. It's now returned to a long term trend and up front purchase price parity has been predicted for mid 2020s for about a decade now. Here an article from 2017: https://about.bnef.com/blog/electric-cars-reach-price-parity-2025/ Anyone working on longer timescales should have been planning for this. China was, the West wasn't and now the cheaper and better cars of the future are Chinese made.

Not yet, NMC Li-ion is still the largest represented in grid scale battery storage [1]. Pumped hydro is dominant overall in storage. LFP represents an increasing share of new grid scale battery storage, though [2] 1. https://www.iea.org/energy-system/electricity/grid-scale-storage 2. https://about.bnef.com/blog/2h-2023-energy-storage-market-outlook/

BloombergNEF reports a cost of $115 per kWh for turnkey energy storage systems (in China) so their comparison is likely to hold up for complete systems in Australia and the US. https://about.bnef.com/blog/global-energy-storage-market-records-bigge…

Like I said, batteries are practical for time shifting a few hours, becauses you can amortize the cost over twenty years and only double the cost of the power, and that is exactly where they are used in grid. https://about.bnef.com/blog/2023-hydrogen-levelized-cost-update-green-… puts cheapest green hydrogen LCoE at $1.50/kg. Hydrogen is scalable to store for long periods in salt caverns and large tanks.

Only because it's a wide line. Cost is $139/kWh, which on a scale of 0-9000, is pretty close to zero historically. https://about.bnef.com/blog/lithium-ion-battery-pack-prices-hit-record…