Sign in

Joe DeCarolis

@jfdecarolis.bsky.social
9.2K followers 252 following 140 posts

Focused on applying state-of-the-art data systems, models, and analysis to inform decision making on energy. Professor at NC State University; served as the 10th Administrator of the U.S. Energy Information Administration. Personal account.

PostsRepliesMedia
Reposted by Joe DeCarolis
Costa Samaras @costasamaras.com · 04/06/2026
Join us on June 10 for a webinar on AI, energy, and U.S. economic growth and inequality.
175
Joe DeCarolis @jfdecarolis.bsky.social · 21/05/2026
Thanks to all the co-authors. This paper was a big team effort. @jxjohnson.bsky.social (NC State) @clameron.bsky.social (Sutubra) Michael Blackhurst (U Pitt) Anderson de Queiroz (NC State) @danielposen.bsky.social (U Toronto) @paulijllo.bsky.social (CMU)
221
Joe DeCarolis @jfdecarolis.bsky.social · 21/05/2026
This analysis assumes data center companies simply bring load to the grid. The price and emissions impacts can be ameliorated if those companies bring their own capacity and/or sign clean energy power purchase agreements.
141
Joe DeCarolis @jfdecarolis.bsky.social · 21/05/2026
The clean energy tax credits (45Y + 48E) in the Inflation Reduction Act would cut the natural gas share from 70% to 41%, with wind (29%) and solar (15%) filling the gap.
101
Joe DeCarolis @jfdecarolis.bsky.social · 21/05/2026
CO₂ rises 9–28% versus a future without data center growth. Mid-demand forecast adds 330 million tons. States with emissions caps (like Virginia under RGGI) import coal power from neighbors, shifting emissions rather than cutting them.
100
Joe DeCarolis @jfdecarolis.bsky.social · 21/05/2026
Costs rise 6–29% nationally by 2030 compared to a future without data center growth. But geography matters. Northern Virginia could see costs jump as much as 57%. Cost burden shifts along with the geographic distribution of demand.
100
Joe DeCarolis @jfdecarolis.bsky.social · 21/05/2026
New construction to meet US data center demand in our mid-range scenario: ☀️ 24 GW solar ⚡ 21 GW natural gas 🌬️ 19 GW wind Gas carries 41–76% of new generation. Coal adds -4% to 23%. Increases mostly associated with aging Midwest and Southeast coal plants running harder.
100
Joe DeCarolis @jfdecarolis.bsky.social · 21/05/2026
We modeled 26 U.S. power regions across 13 scenarios varying demand, incentives, natural gas prices, gas turbine costs, and the distribution of data center demand. Data center demand ranges from 10–19% of total U.S. electricity in 2030. We also ran a counterfactual without data center demand.
110
Joe DeCarolis @jfdecarolis.bsky.social · 21/05/2026
🔌💡 New paper alert: Surging data center and cryptocurrency demand could spike regional electricity costs up to 57% and push U.S. power sector CO₂ emissions up 28% by 2030. iopscience.iop.org/article/10.1... A thread on what we found 🧵👇
iopscience.iop.org
Power system costs and emissions from data center and cryptocurrency mining expansion in the United States
Power system costs and emissions from data center and cryptocurrency mining expansion in the United States, Johnson, Jeremiah X., Wade, Cameron, Blackhurst, Michael, Decarolis, Joseph F, De Queiroz, Anderson R, Posen, I Daniel, Jaramillo, Paulina
11611
Joe DeCarolis @jfdecarolis.bsky.social · 18/05/2026
Note that we created a similar open-source dashboard at EIA, available here: github.com/EIAgov/dash-.... The underlying data includes the Reference case and as many of the side cases as we could scrape together.
github.com
GitHub - EIAgov/dash-benchmark
Contribute to EIAgov/dash-benchmark development by creating an account on GitHub.
130
Joe DeCarolis @jfdecarolis.bsky.social · 30/04/2026
Thanks Doug, much appreciated!
010
Joe DeCarolis @jfdecarolis.bsky.social · 09/03/2026
Brent and WTI above $100. Strait of Hormuz largely shut down. EIA has the data and expertise to explain what might happen next. They've been quiet so far, but hopefully tomorrow's Short-Term Energy Outlook will offer some insight.
081
Joe DeCarolis @jfdecarolis.bsky.social · 13/02/2026
Obviously terrible BUT I keep reminding myself that what has been undone can be redone. The science doesn't change, and eventually... EVENTUALLY... we will realign U.S. policy with the science.
081
Joe DeCarolis @jfdecarolis.bsky.social · 23/01/2026
FYI , the original source of that figure was a 2016 EIA report with the (wordy) title: "Wind and Solar Data and Projections from the U.S. Energy Information Administration: Past Performance and Ongoing Enhancements" www.eia.gov/outlooks/aeo...
eia.gov
120
Joe DeCarolis @jfdecarolis.bsky.social · 17/01/2026
This animation looks similar to the ones I developed and presented at MIT in 2024 when I was EIA Administrator: www.youtube.com/watch?v=V0fL... My explanation for natural gas production begins around the 25:50 mark, followed by crude oil, wind, solar, and CO2 emissions.
youtube.com
Stay humble and prepare for surprises: Lessons for the energy transition with Joseph F. DeCarolis
YouTube video by MIT Energy Initiative
161
Joe DeCarolis @jfdecarolis.bsky.social · 03/12/2025
Yes, the economics are pretty simple: incentivize demand for gasoline and prices will go up. Look no further than EPA's own regulatory impact analysis earlier this year, which cited EIA projections: www.cbsnews.com/news/epa-rul...
cbsnews.com
Repealing rule to curb greenhouse gas emissions will increase gas prices, Trump administration's own analysis finds
EPA Administrator Lee Zeldin told CBS News' "The Takeout" that environmental policy can't "strangulate out of existence" energy policy.
12414
Joe DeCarolis @jfdecarolis.bsky.social · 13/11/2025
Funny story about George Dantzig's formulation of the canonical diet program to test his newly developed simplex algorithm. The lesson? Embed some common sense in the model formulation. Thanks to @bistline.bsky.social for sharing.
0101
Reposted by Joe DeCarolis
Jeremy Michalek @jmichalek.bsky.social · 22/09/2025
Today we submitted our public comment on critical flaws in EPA's rationale for rescinding its authority to regulate greenhouse gas emissions. This proposed move is a big deal with profound implications for our future. Today's the last day to comment. www.cmu.edu/cit/veg/publ...
25217
Joe DeCarolis @jfdecarolis.bsky.social · 07/08/2025
Excellent piece by @bencasselman.bsky.social on the BLS data revisions. The U.S. Energy Information Administration, another federal statistical agency, faces similar data challenges. The accuracy vs timeliness tradeoff is universal: data collected at higher frequency is inherently noisier.
0153
Reposted by Joe DeCarolis
Georg Zachmann @gzachmann.bsky.social · 22/06/2025
And here the timely EIA data (and analysis) on oil trade volumes through the Strait of Hormuz - 20% of global oil consumption: www.eia.gov/todayinenerg...
141
Joe DeCarolis @jfdecarolis.bsky.social · 13/05/2025
I was happy to participate in this event and explain how the U.S. Energy Information Administration meets energy data needs in the United States. 🔌💡
081
Joe DeCarolis @jfdecarolis.bsky.social · 06/05/2025
Some personal news: This November, I'll be returning to Carnegie Mellon (@cmu.edu) as head of the Department of Engineering and Public Policy (EPP). I'm honored and excited to help lead EPP into its next chapter. engineering.cmu.edu/news-events/...
engineering.cmu.edu
DeCarolis named head of Engineering and Public Policy
Joseph DeCarolis has been named head of the Department of Engineering and Public Policy.
7384
Joe DeCarolis @jfdecarolis.bsky.social · 05/05/2025
Wind is hard to model in capacity expansion models because its performance is highly dependent on its spatiotemporal representation. NEMS, like many other models, has a relatively stylized representation of wind resources.
020
Reposted by Joe DeCarolis
Rep. Sean Casten @casten.house.gov · 30/04/2025
If you want to lower energy costs for Americans, you must first understand the US energy system. You can't do that without the robust, accurate data the EIA provides. The Trump Administration's haphazard dismantling of the EIA will hurt our ability to increase access to cheap, reliable energy.
reuters.com
Over 100 employees leave US EIA, putting crucial energy data at risk, sources say
The U.S. government's energy statistics arm is set to lose over 100 employees after the Trump administration's latest round of resignation offers, putting at risk some of the most closely watched energy reports globally, three sources told Reuters.
912331
Joe DeCarolis @jfdecarolis.bsky.social · 18/04/2025
Last year we also released an open source python dashboard that enables comparisons across AEO editions: github.com/EIAgov/dash-... (Looks like it hasn't been updated with AEO2025 data yet.)
github.com
GitHub - EIAgov/dash-benchmark
Contribute to EIAgov/dash-benchmark development by creating an account on GitHub.
010
Joe DeCarolis @jfdecarolis.bsky.social · 18/04/2025
All credit goes to EIA staff. (I departed EIA on Jan. 20.) The EIA Retrospective Report, released every other year, calculates statistics across previous AEO editions: www.eia.gov/outlooks/aeo...
100
Joe DeCarolis @jfdecarolis.bsky.social · 17/04/2025
That's all for now. Again, I appreciate the dedication of EIA staff to produce the AEO under challenging circumstances. And remember EIA's transparency. Don't like a result? Run NEMS with your own assumptions: github.com/EIAgov/NEMS
github.com
GitHub - EIAgov/NEMS
Contribute to EIAgov/NEMS development by creating an account on GitHub.
010
Joe DeCarolis @jfdecarolis.bsky.social · 17/04/2025
If you’re curious about the Trump Administration’s reaction to these independently produced results: www.energy.gov/articles/doe...
energy.gov
DOE Statement on EIA Annual Energy Outlook
The Department of Energy (DOE) today released the following statement from DOE Spokesperson Andrea Woods on the U.S. Energy Information Administration (EIA) Annual Energy Outlook 2025
100
Joe DeCarolis @jfdecarolis.bsky.social · 17/04/2025
That’s a brief overview of some key results. Remember, modeling assumptions were frozen in December 2024. The effects of Trump’s executive orders and tariff policy, among other factors, could have a significant impact on these projections.
100
Joe DeCarolis @jfdecarolis.bsky.social · 17/04/2025
EV sales also hit a hard upper limit around 55% of new sales. I suspect this has to do with the relative economics between the different vehicle classes as well as details in the consumer choice model. In future AEO editions, EIA should explore assumptions that could lead to higher EV deployments.
110
Joe DeCarolis @jfdecarolis.bsky.social · 17/04/2025
Finally, here’s a look at electric vehicle sales. Excluding several recent transportation-related regulations has a big impact on electric vehicle (EV) deployment.
Electric vehicle share of new vehicle sales from EIA's Annual Energy Outloook 2025.
100
Joe DeCarolis @jfdecarolis.bsky.social · 17/04/2025
You can see from the figure above that electrified transport has a significant impact on electricity demand – note the difference between the Reference case and Alternative Transportation case.
100
Joe DeCarolis @jfdecarolis.bsky.social · 17/04/2025
Overall, total electricity generation increases nearly 50% from 2024-2050. This increase is due to the electrification of end-uses, like transportation, as well as increases in existing sources of electricity demand, like data centers.
Total electricity generation from EIA's Annual Energy Outlook 2025.
183
Joe DeCarolis @jfdecarolis.bsky.social · 17/04/2025
And finally, here’s generation from natural gas, which declines over time in most cases as renewables take up an increasing share of generation.
Natural gas-fired generation projections from EIA's Annual Energy Outlook 2025.
110
Joe DeCarolis @jfdecarolis.bsky.social · 17/04/2025
Here’s nuclear generation. Consistent with past AEO editions, nuclear energy remains relatively flat but with some variation. For example, the Low Oil and Gas Supply case leads to more expensive gas, which incentivizes new nuclear capacity in later years.
Nuclear generation projections from EIA's Annual Energy Outlook 2025.
100
Joe DeCarolis @jfdecarolis.bsky.social · 17/04/2025
Here’s coal. You can see the effect of the EPA 111d regulations on coal generation – it forces nearly full coal retirement by 2040. Note: “Alternative Electricity” removes those new regulations.
Coal-fired electricity generation projections from EIA's Annual Energy Outlook 2025.
100
Joe DeCarolis @jfdecarolis.bsky.social · 17/04/2025
Let’s step through the full set of projections for each major source of electricity generation. Here’s renewables. (Unfortunately, wind and solar are lumped together in the tables.) Note the marked increases in renewable generation and slowdown in later years when the IRA tax credits expire.
Renewables (wind + solar) generation from EIA's Annual Energy Outlook 2025.
200
Joe DeCarolis @jfdecarolis.bsky.social · 17/04/2025
Here’s a view into the electric sector, with the AEO2023 Reference case, current Reference case, and Alternative Electricity case. Note the consistently high renewables generation, between 60-70% across these three cases.
Electricity generation by source from EIA's Annual Energy Outlook 2025
100
Joe DeCarolis @jfdecarolis.bsky.social · 17/04/2025
So as renewable penetration increases and end uses are electrified, all else equal, it decreases the need for thermal-based primary energy sources.
110
Joe DeCarolis @jfdecarolis.bsky.social · 17/04/2025
EIA changed its primary energy accounting for noncombustible renewables: www.eia.gov/totalenergy/... In AEO2025, 1 unit of solar or wind electricity = 1 unit of primary energy. So 1 unit of solar or wind displacing 1 unit of fossil electricity reduces fossil primary energy by 2-3 units (2nd Law).
eia.gov
U.S. Energy Information Administration - EIA - Independent Statistics and AnalysisU.S. Energy Information Administration (EIA) - Pub
100
Joe DeCarolis @jfdecarolis.bsky.social · 17/04/2025
Here's U.S. primary energy consumption. You can see substantial reductions from AEO2023 and through 2040. What’s going on? Renewables, mostly solar and wind, are an increasing share of the electricity mix. End-use electrification is also increasing. What does that have to do with primary energy?
Total primary energy from EIA's Annual Energy Outlook 2025.
100
Joe DeCarolis @jfdecarolis.bsky.social · 17/04/2025
The 2050 Reference case emissions are 17% lower in AEO2025 than AEO2023. It’s largely due to the IRA, EPA regulations, and continued cost declines in renewables and batteries. Emissions tend to flat-line around 2040 as IRA credits expire and growing energy demands exert influence.
100
Joe DeCarolis @jfdecarolis.bsky.social · 17/04/2025
OK, now let’s dig into the results, starting with energy-related CO2 emissions.
Energy-related CO2 emissions from EIA's Annual Energy Outlook 2025.
100
Joe DeCarolis @jfdecarolis.bsky.social · 17/04/2025
Each figure includes the results across different cases. Insights should always be based on carefully examining the full range of outcomes. You can’t really learn anything useful from a single projection.
110
Joe DeCarolis @jfdecarolis.bsky.social · 17/04/2025
In the charts below, the Reference case is the bold black line. Do NOT interpret the Reference case as a forecast. Think of it as an experimental control – it’s what happens under nominal conditions. Read the Administrator’s Forward from AEO2023 if you want my full take: www.eia.gov/outlooks/aeo...
eia.gov
Narrative - U.S. Energy Information Administration (EIA)
100
Joe DeCarolis @jfdecarolis.bsky.social · 17/04/2025
Compared with AEO2023, this latest edition includes the major changes to NEMS referenced above, updates to end-use demands and technology cost projections, and changes in policy, such as EPA’s April 2024 revisions to CO2 standards under Section 111 of the CAA.
100
Joe DeCarolis @jfdecarolis.bsky.social · 17/04/2025
I’ve included the AEO2023 Reference case in the charts since it’s been two years since the last AEO was released. When I was Administrator, I made the decision to focus on model updates in 2024 and skip AEO2024: www.eia.gov/outlooks/aeo...
eia.gov
Annual Energy Outlook 2025 Model Development - U.S. Energy Information Administration (EIA)
100
Joe DeCarolis @jfdecarolis.bsky.social · 17/04/2025
In addition to the Reference case and core side cases that EIA typically runs, they also included separate “Alternative Electricity” and “Alternate Transportation” cases, which remove some Biden-era regulations. They provide a useful point of comparison.
100
Joe DeCarolis @jfdecarolis.bsky.social · 17/04/2025
The modeled cases in AEO2025 are described here: www.eia.gov/outlooks/aeo...
eia.gov
U.S. Energy Information Administration - EIA - Independent Statistics and AnalysisEIA - Annual Energy Outlook 2023 Case DescriptionsU.S. Energy Information Administration (EIA) - Source
100
Joe DeCarolis @jfdecarolis.bsky.social · 17/04/2025
Keep in mind that the National Energy Modeling System (NEMS), used to produce the AEO, is complex. Assumptions were frozen in December 2024 in order to complete the model runs. So the results reflect an uncertain world with Biden-era policies fully intact.
100