Steel For Fuel N° 051 of 56 21 Apr 2026
Be greedy (for electrons) when others are fearful
by Andy Lubershane, Partner and Head of Research, Energy Impact Partners
In this note
The question
With American electric vehicle sales falling sharply after the federal tax credits expired, is this the moment to retreat from the electric vehicle supply chain and from charging?
The answer
No, the reverse. Autonomous vehicles are nearly always electric for cost reasons rather than emissions reasons, and Lubershane argues they are at an inflection point, so the consumer market’s contraction is the wrong thing to read. What he points at is charging infrastructure rather than a large jump in electricity demand.
03The argument
The post opens with the case for fear stated in full: credits gone, plug-in sales roughly halved, tens of billions written off by American automakers, and the country overtaken by China, Europe and now emerging markets. He says he understands the impulse, then treats it as the signal to buy. His reason is one he says he has made before and quotes from his own earlier writing. Autonomous vehicles will be high-utilization vehicles, because they can be and because they must be to justify their capital cost, and high utilization is the regime in which an electric drivetrain’s lower maintenance and energy costs, which he qualifies as holding in most places, outweigh its purchase price. The sensors and processors must be powered regardless, which compounds the case. Growth in autonomy is therefore growth in demand for electricity.
The extension past passenger cars is where the argument turns. Heavy trucks have resisted electrification for reasons he sets out plainly: battery weight against range, sticker price, the cost of installing chargers, and reluctance to adopt an unfamiliar system, all despite ownership costs already favouring them in some duty cycles. Against that he sets two things, that most American truck freight travels short distances well inside current electric range and that plug-in trucks already hold a large share of the Chinese heavy duty market, and concludes that autonomy will be the spark carrying American operators over a threshold cost alone has not. He is careful about the size of the prize: the added electricity demand is a fraction of a percentage point, which he concedes is small next to hyperscale data centers. The opportunity is the charging infrastructure, several gigawatts of it, much of it in depots where vehicles would be charged, cleaned and serviced together.
04What you need to know first
- The autonomy stack
- The cameras, lidar, radar and processors a self-driving vehicle carries. They draw power continuously.
- Utilization
- The share of time a vehicle is in service. The higher it is, the more running costs matter against purchase price, which is the pivot of the cost argument.
05Details worth keeping
- The epigraph is an investing maxim the post attributes to “Warren Buffet”.
- On the state of the field: Waymo is on the march, Zoox is reportedly testing in ten American cities, and Tesla is still trying.
- His back-of-the-envelope analysis, he says, already makes the unit economics of autonomous electric vehicles attractive against human rideshare drivers, with room left to cut hardware cost; the comparison itself appears only as an unlabelled chart, so this note cannot say what it showed.
- Terawatt and Voltera are named as companies with a head start on depot-style charging sites.
- The Chinese electric truck surge, he notes, came before the closure of what he spells the “Straight of Hormuz”, without saying what he draws from it.
06Claims worth citing
All figures as stated on 2026-04-21. Sales and write-off figures move quickly, and the demand estimates are his own arithmetic rather than a forecast.
- Sales of both fully electric vehicles and plug-in hybrids are down by about 50% since the tax credits expired. Lubershane, from an Alliance for Automotive Innovation dashboard
- The three large American automakers have written off about $53 billion in electric vehicle supply chain investments, and Honda a further $16 billion. Lubershane
- Computing alone could account for about 45% of an autonomous vehicle’s energy, though most estimates he has seen put it at 20 to 30% for urban driving, comparable to a few NVIDIA H100 processors at 700 watts each running full tilt. Bloomberg for the first figure, unnamed sources for the range, cited by Lubershane
- More than 85% of American truck freight travels less than 250 miles and more than 40% less than 100 miles, both inside current electric truck specifications. Bureau of Transportation Statistics, cited by Lubershane
- Trucks with a plug took nearly a third of China’s heavy duty market last year, crossing 50% as sales surged in the final months. Electrive and the International Council for Clean Transportation, cited by Lubershane
- Robotaxis capturing 1% of the 2.9 trillion miles American passenger vehicles travelled in 2024 would raise electricity demand about a third of a percentage point and, if most charging is done slowly and overnight, need nearly 4 gigawatts of new charging. Lubershane’s own calculation, assumptions in his footnotes
07Where it’s contested
Nothing is disputed here; there is no second voice. The post carries marked confidence levels and one large undefended assumption.
- The block-quoted passage is his own earlier writing, introduced as something he is repeating rather than revising. The case for why autonomous vehicles are electric is quoted from a previous post.
- The load-bearing assumption is that autonomy scales. Everything follows from exponential growth over a decade or more, stated as conviction and supported by his own back-of-the-envelope unit economics and the progress of three named companies rather than tested.
- The answer to the truck objection is thinner than the objection. He sets out carefully why operators have balked, then answers that autonomy will be the spark, without saying how it addresses sticker price or charger cost.
- He hedges the sizing, opening the estimate by saying nobody knows and calling it basic math.
- What he has at stake. The post announces his firm’s investment in Humble, described by its concept rather than by any performance figure, and ends by asking readers who agree to get in touch.