Home / Insights

Field guide

The engine is near its efficiency ceiling. The battery's headroom is cost.

Forty years of EPA ratings show gasoline efficiency has stalled and hybrids have plateaued. Electric drive is already efficient. What is left to win in batteries is cost and energy density.

By Dan Lee · November 2026

The most fuel-efficient gasoline car you could buy in 1986 was rated 48 mpg. The best one on sale for 2026 is rated 36. The best hybrid has been rated between 50 and 59 mpg for 26 years. Since 2012, the best battery-electric car rose from 112 to 146 mpge. Combustion has run into its ceiling. Electric drive is close to its ceiling too, but from a much higher floor. The room left in batteries is not efficiency. It is cost and energy density.

In brief

• EPA ratings for every vehicle sold since 1984 show the gasoline frontier flat to down, and the typical gasoline model up only from 19 to 21 mpg.
• Engines did improve a lot. Fuel burned per horsepower is down more than 70% since 1975. Buyers spent the gain on power, weight and trucks.
• The shift from sedans to SUVs and pickups costs today's fleet about 4.8 mpg. Within each segment, the typical gasoline model has been flat since 2016, and hybrids help pickups and large SUVs the least.
• Physics caps the engine. Only 18% to 25% of the fuel's energy reaches the wheels. The best production engines peak near 41% to 42% thermal efficiency, and the last big gain, the hybrid, has also plateaued.
• Electric drive already sends more than 77% of grid energy to the wheels. Efficiency gains from here are single digits a year at best.
• The headroom is in the battery: pack prices fell to $108/kWh in 2025 and keep falling about 20% per doubling of production, and cell energy density has a credible path from about 300 to 400 Wh/kg.

Forty years of EPA ratings, by powertrain

The EPA has rated every new light vehicle sold in the U.S. since 1984, and publishes the full list. We sorted it by powertrain: gasoline and diesel, strong hybrid, plug-in hybrid and battery electric. For each model year we took the best-rated model and the median model.

Three patterns stand out.

Gasoline is flat. The best non-hybrid peaked in the late 1980s with three-cylinder minicars near 1,600 pounds. Cars that light no longer pass today's crash standards, so the frontier fell to the high 30s and stayed there. The median gasoline model moved from 19 mpg to 21 in four decades.

Hybrids made one big step, then stopped. The first U.S. hybrid in 2000 rated 53 mpg. The best hybrid today rates 57. Hybrids moved from a novelty to 9.5% of production, but the top of the range has not moved since 2016.

Electric is high and rising slowly. The best electric car gained about 30% in 13 years, roughly 2% a year, from aerodynamics, better motors and inverters, and heat pumps. The median electric model has fallen since 2020, for a reason the segment data makes clear.

Where the engine's gains went

The flat ratings do not mean engineers stood still. The EPA's own Automotive Trends Report shows the opposite: fuel consumed per horsepower is down more than 70% since 1975, and the gain in power per liter of displacement "does not appear to be slowing." Direct injection is now on almost 80% of new vehicles. Stop/start is on three quarters.

Buyers spent most of it. Since 1980 average horsepower is up 154% and weight is up 38%. Zero to 60 fell from 15.6 seconds to 7.3. Trucks and SUVs went from 17% of production to 65%. Real-world fuel economy rose 46%, and part of the recent climb comes from electric vehicles entering the fleet average, not from better engines.

That is a choice, and choices can change. But it tells you how much is left. Decades of engine work bought large gains in what the engine could do and modest gains in what it burned.

The mix is pulling the average down

A large drag on the fleet average is what Americans buy. Sedans and wagons were 84% of new-vehicle production in 1980 and 23% in 2025. SUVs regulated as trucks went from 2% to 48%. Pickups held at about 15%.

Every vehicle type got more efficient. Since 2005, the EPA's real-world average rose 54% for sedans and wagons, 57% for truck-based SUVs and 35% for pickups. But if 2025's vehicles had been sold in the 1980 mix, the fleet would average 32.9 mpg instead of 28.1. The shift to bigger vehicles costs about 4.8 mpg, roughly 15%. Even against the 2005 mix, it costs 2 mpg.

Those type averages include the hybrids and EVs sold in each type, so they flatter the engine. Cutting the EPA ratings by segment and powertrain shows what the engine alone has done.

Three things stand out.

• Within every segment, the typical gasoline model has stopped improving. The 2026 medians are 23 mpg for cars, 24 for crossovers, 19 for standard SUVs, 19 for pickups and 22 for minivans. Each has been about flat since 2016.
• Electric drive is about four times as efficient in every segment. Median electric ratings run 93 mpge for cars, 103 for crossovers, 86 for SUVs, 78 for pickups and 82 for minivans. The ratio holds whatever the body style.
• Hybrids help least where the volume is. A hybrid doubles a car's rating (46 against 23) and adds about 60% to a crossover or minivan. It adds about 40% to a standard SUV and about 20% to a pickup (23 against 19). Heavy vehicles that spend their time at highway speed or towing give the hybrid less braking energy to recover. That matters, because trucks and SUVs are now most of the market.

Commercial vans tell the same story. Full-size work vans weigh more than 8,500 pounds gross, so they carry no fuel-economy label. Most are still tested in the EPA's lab for emissions certification, and those public test results show the gasoline van stuck near 15 to 18 mpg and the diesel van near 17 to 21 since 2014, before any real-world adjustment. Electric vans use roughly a quarter to a third of the energy. The 2025 electric Transit covers up to 159 miles on an 89 kWh pack, about 56 kWh per 100 miles, against roughly 200 to 280 kWh-equivalent for a gasoline van depending on how the lab figure is adjusted. Vans are where electric drive pays off fastest, because their routes are short, predictable and end at a depot charger.

The mix works inside electric vehicles too. The median electric model fell from 111 mpge in 2020 to 88 in 2025 because the electric lineup moved from small cars into SUVs and pickups. The cheapest efficiency gain left in the fleet is buyers choosing smaller vehicles, and nothing in four decades of data says they will.

Why the ceiling is real

A gasoline vehicle delivers 18% to 25% of its fuel's energy to the wheels in combined driving, according to the Department of Energy. Most of the rest leaves as heat. That loss is thermodynamics, not engineering neglect.

The best production engines reach about 41% to 42% thermal efficiency at their single best operating point. Nissan has shown 50% in the lab, but only by running the engine at a fixed speed and load with waste-heat recovery. A fixed-speed engine is a generator. In other words, the path to a better engine runs through electrification.

That is what the hybrid does. It lets the engine run near its sweet spot and recovers braking energy. In 2024 production, strong hybrids used about 25% less energy per mile than conventional gasoline vehicles, in EPA's real-world estimates. It is the largest single efficiency step combustion has left, and it has already been taken by the best hybrids.

Electric drive is efficient already

Electric vehicles convert more than 77% of the electricity drawn from the grid into motion at the wheels, by the Department of Energy's estimate. Motors and inverters are above 90% efficient on their own. The average 2024 electric vehicle used about 34 kWh per 100 miles, a quarter of the energy of a conventional gasoline car.

So the honest version of the battery story is not "electric vehicles will keep getting more efficient." They will, a little. The best ratings rise about 2% a year, and the median can fall when buyers choose larger vehicles. Physics caps an electric motor too. It just caps it much higher.

The battery's headroom is cost and density

Where batteries still have room is what they cost and how much energy they hold per kilogram.

Cost. BloombergNEF's 2025 survey put the average lithium-ion pack at $108/kWh, down 8% in a year, with packs for electric cars at $99. Lower-cost LFP packs averaged $81, against $128 for nickel-rich NMC. China averaged $84, while North America ran 44% higher. As we showed in our piece on the cost curve of new technology, cell prices have fallen about 20% for every doubling of cumulative production. The industry keeps doubling.

Density. Production NMC cells hold roughly 250 to 300 Wh/kg, and LFP 150 to 180. Solid-state developers report sample cells near 375 Wh/kg; Factorial and Stellantis validated automotive-size cells at that level in 2025. Those figures are self-reported and not yet in volume production, but the direction is clear. A 30% gain in density is the kind of improvement combustion has not offered in decades.

Density also feeds back into efficiency. Battery weight is the main reason heavy electric trucks rate poorly. A lighter pack for the same range means a lighter vehicle and better mpge, and a cheaper pack means automakers can offer range without the weight penalty at the top of the lineup.

What this means for owners and investors

• Combustion suppliers. The efficiency content story is mostly over. Direct injection, turbochargers and 8-to-10-speed transmissions are already on most vehicles. Growth comes from hybrid content (power electronics, e-motors, thermal management), not from the next turn of the engine. Value the core business on cash flow, not technology upside.
• Hybrid exposure. The hybrid is combustion's last big step and the bridge most automakers are now leaning on. Suppliers positioned on both sides of it carry less risk than either pure play.
• Battery and EV suppliers. Expect your selling price per kWh to fall every year. Cost curves reward the producer with scale and punish the one that prices off today's cost. Plan margins against a moving target.
• Diligence questions. For any business plan that assumes efficiency gains, ask which technology is producing them and how close it sits to its physical limit. For any plan that assumes battery cost, ask which chemistry, which region and which point on the cost curve.

How we read the data

The EPA rating series counts each vehicle configuration once, regardless of how many were sold, so it shows what was available, not what was bought. The sales-weighted figures come from EPA's Automotive Trends Report, whose 2025 values are preliminary. Window-sticker ratings understate real-world consumption for most drivers; the trends report adjusts for that. Plug-in hybrid results depend heavily on how often owners charge. We count mild hybrids (48-volt systems) separately from strong hybrids, because they behave like conventional vehicles. Segments follow EPA size classes: the EPA rates small SUVs (crossovers) separately only from 2013, and its trends report splits SUVs by how they are regulated, not by body style. Commercial vans are not in the label data, because most weigh more than 8,500 pounds gross. Our van figures come instead from EPA certification lab tests (city and highway cycles, converted from CO2), which are unadjusted and run about 20% to 30% better than real-world results; only 2 to 9 vans are tested each year, and models carry over between years. Electric van figures are manufacturer range and pack size, not lab energy tests. Minivans have only a handful of rated configurations each year.