MIT: Zero U.S. ZIP Codes Where Gas Beats EVs on Emissions

The argument had one job: find a single American ZIP code where a gas car beats an EV on lifetime greenhouse gas emissions. MIT looked at all of them. There isn't one.

That is the finding from the MIT-led lifecycle analysis making the rounds this week, and it closes the last empirical escape hatch for the dirty-grid objection. The efficiency gap between a battery-electric drivetrain and an internal-combustion one is wide enough that even the coal-heaviest corner of the American grid cannot flip the outcome over a vehicle's service life.

Zero ZIP Codes. That's the Number.

The methodological move is the story. Previous lifecycle studies compared national or state averages, which left room to argue that some specific location, your location, presumably, was the exception. MIT ran the numbers at ZIP-code resolution against the actual generation mix serving that ZIP. The result: no location produced higher lifetime EV emissions than a comparable gas car.

Manufacturing is where the gas-car side of the ledger looks best. Building an EV emits roughly 50–80% more CO₂ upfront than building a comparable gas car, and the electricity used to charge it carries carbon that varies with how the local grid is generated, coal and natural gas at one end, wind and solar at the other. Coal-heavy grids slow the payback window. They do not reverse it. The efficiency of an electric drivetrain, roughly three to four times the well-to-wheel efficiency of combustion, compounds every mile until the manufacturing debt clears and then keeps compounding.

Cold-climate penalty exists and is quantified in the same analysis. Range drops, heat-pump losses stack, and the payback stretches. It still doesn't close the gap. An MIT-led study found EVs usually beat gas cars on emissions, but the advantage varies sharply by location and driving habits, "varies sharply" is doing real work in that sentence, but the variance is between how much cleaner, not whether.

The framing that EVs simply offload tailpipe emissions to the power plant has been the load-bearing beam of anti-EV argument for a decade. MIT's ZIP-code framing closes the location escape hatch the argument depended on.

Canada's Grid Makes the Math More Lopsided, Not Less

Canada's national grid runs roughly 83% non-emitting, hydro, nuclear, wind, and a growing solar contribution. The manufacturing debt clears faster here than in most American states, and the lifetime delta widens correspondingly. In British Columbia and Quebec, where hydro dominates, lifetime EV emissions land in near-zero territory on a well-to-wheel basis.

Alberta and Saskatchewan are the stress cases. Their grids still lean heavily on natural gas and, in Alberta's case, residual coal. Even there, the MIT framework holds directionally, the payback window is longer, not infinite. The gap the study measured in West Virginia is wider than anything in the Canadian mix.

EVAP compresses the upfront-cost side of the equation MIT treats as fixed. A $5,000 federal rebate against a BEV under $50,000 shifts the total-cost-of-ownership crossover forward, and the emissions crossover was already sitting somewhere between 15,000 and 40,000 kilometres depending on the grid. The efficiency argument that Chinese programmes prioritised a decade ago, shape and drag over battery size, is the same argument MIT is measuring here, in different clothes. For readers weighing brand strategy against grid math, the technical breakdown of why shape beats curb weight sits in the same family of arguments.

The Dirty-Grid Objection Was Always a Delay Tactic

The dirty-grid argument was structurally unfalsifiable by design. Every grid improvement was met with a new threshold: fine, but what about coal states; fine, but what about winter; fine, but what about a specific ZIP code somewhere in Appalachia. MIT's ZIP-code framing removes the last place the objection could hide, because the study checked every location.

The next iteration of the argument will move to battery mining, and the emissions-accounting framework MIT built applies there too, the numbers are already in the lifecycle total. The empirical debate is closed. The political one isn't, and from here on, conflating the two is the tell.