Right now millions on the East Coast are getting a preview of what breathing bad air feels like. A new study looks at who breathes it by design. Net-zero pathways with more carbon removal leave ~30,000 more Americans dying from PM2.5 in 2050, concentrated in lower-income places.

Jevons strikes again: new paper on 30,000+ Chinese firms shows that adopting AI cuts a company’s carbon intensity 6.8% on average. Rebound effect eats 70% of it, so actual emissions only fall 2%.

Every corporate sustainability team is asking the same question: How do you measure AI’s carbon footprint? Spent months on this with AI providers and academic collaborators. Walking through an open accounting framework on July 23, live.

Everyone remembers Cash for Clunkers as a car sales program. It was really a pickup and SUV disposal program. 66% of everything destroyed was a Category 1 truck.

Gas prices jumped from $3.00 to $4.50 a gallon this year. Americans responded by buying more hybrids, not more EVs. New RFF analysis on why the “obvious” substitution didn’t happen.

Corporate clean power buying fell in 2025 for the first time in a decade. Our customers at Watershed went the other direction: $500M and counting, funded by pooling demand from companies too small to do these deals alone.

Housing costs rose $6,220/year since 2019. Electricity rose $480. Yet electricity is the one households keep hearing about. New LBNL data from Ryan Wiser’s team: electricity is 2.5% of the total increase in what Americans spend each year.

EVs are the marquee climate subsidy. They also have one of the lowest returns per dollar spent. New MVPF analysis: wind production tax credits deliver $5.87 in social benefit per $1 spent. Residential solar: $3.86. EV subsidies: $1.45.

Forget heat pumps. California’s central coast solved building energy use by being next to cold ocean water. Almost zero heating demand and almost zero cooling demand all year. Gotta upzone Big Sur.

The federal government is now paying utilities to keep coal plants open past retirement. This paper’s evidence says the people living near those plants are the ones paying twice: once in their utility bill, once by moving away from the pollution.

New MIT modeling: Over two-thirds of the clean electricity generation and emissions cuts under Biden’s climate law could survive under Trump’s rewrite moving forward. But falling costs aren’t a guarantee of future progress: permitting, fuel costs, and demand all matter.

De minimis for most U.S. net-zero scenarios: www.sciencedirect.com/science/arti...

Hormuz traffic is finally ticking up. 7-day average back to ~33/day after the June 17 MOU. Still about a third of normal for this time of year.

I’m aiming to get the code up soon (troubleshooting an API issue first). In the meantime, I created California a few weeks ago (before I adjusted the color palette to be color-blind-friendly).

Texas cuts against stereotypes of people outside energy. Although Texas is a leading fossil fuel producer, it is also a renewables and energy storage powerhouse. Very difference from, say, France.

In a similar spirit, I want to create a visualization to show the solution arriving. So I built Clean Energy Stripes. For an inaugural installment, here’s a grid that surprises many people: Texas.

Climate scientist Ed Hawkins created warming stripes to make climate change and temperature trends feel more intuitive and local.

Texas has no clean energy mandate. It just beat fossil fuels with wind, solar, nuclear, and hydro on 62% of days this year. Nobody told the grid to do this, just the market.

AC debates assume universal adoption is a climate villain. New PBL/IMAGE modeling: closing the global cooling gap entirely (3.6 billion people!) raises 2050 emissions by just 3.6%. Universal cooling’s climate footprint is… small.

15 years ago today: Michele Bachmann channeled John Wayne’s spirit in Waterloo, Iowa. Factually accurate, just not the John Wayne she had in mind.

The causal estimates: a data center doubling lowered residential rates by ~4%. The naive OLS is near zero; adding the instrument shifts it clearly negative. Robust to dropping California, dropping 2020, etc. Near-zero effects for commercial and industrial prices.

Fiber was largely laid along interstate corridors, and data centers follow. So the 1947 plan (based on population, defense, bison trails), predicts modern data centers. Conditional on population/GDP, that 80-year-old map doesn't affect today's rates except through siting.

But maybe data centers lower overall costs but shift them to residential ratepayers. The table rejects this. In both between- and within-state comparisons, higher C&I demand is associated with lower residential prices (negative coefficient). No systematic cost-shifting.

A key finding hiding in cost data: post-2015 increases were driven by T&D capital spending, which was up ~80% from 2015. Generation opex trended down for most of the decade. If your bill went up, blame wires and infrastructure investment.

We test this directly using utility-level cost data (FERC Form 1). Across all four major cost categories (generation capex/opex, T&D capex/opex), estimates are below 1. That means a 1% increase in demand is associated with <1% cost increase: economies of scale.

"But doesn't more demand raise prices?" Not necessarily in electricity. Retail rates recover huge fixed costs spread across every kWh. New durable demand can spread those fixed costs over more kWh.

Simple picture: Plot each state's data center growth against its price change, and there's essentially no positive relationship. VA, the U.S. data center leader (~20% of its power), had near-average rate increases. CA's 40% spike was wildfire costs rather than data centers.

That’s a good one!

One of my favorite parts of Toy Story is the Dinoco gas station, which is reminiscent of the Union 76 gas in Beverly Hills. Are there other examples of midcentury modern energy infrastructure?