Car companies have struggled to sell electric cars. Tesla Motors is the exception. Last year, the first full year of sales for its Model S luxury sedan, Tesla sold more than twice as many cars as either Nissan or GM did when they introduced their battery-powered vehicles, the Leaf and the Volt. Tesla did this even though it’s a startup with no dealer network, selling a car that’s more than twice as expensive as the electric cars from the major automakers.

It’s easy to be dazzled by the car’s style or features like its 17-inch touch screen. But the innovation goes much deeper than that. MIT Technology Review’s senior editor for energy, Kevin Bullis, asked JB Straubel, Tesla’s cofounder and chief technical officer, to help identify the engineering advances behind Tesla’s success.

Other electric-car startups—such as Fisker Automotive—have failed. And even established automakers have struggled to sell their battery-powered cars. What makes you different?

Part of it is that Tesla designs its own batteries, motors, electronics, and software controls. It’s not very glamorous, and not even always customer facing, but in the end that’s what makes the car work and what makes it different from other electric cars, and makes it compete effectively against gasoline cars.

An example?

Our superchargers allow us to charge the Model S more than twice as fast as other cars. To do that kind of charging, everything has to be working in perfect synchrony. The cooling system; the electronics that are talking to the charger; the connection to the grid. That whole thing has to work as a system flawlessly. If we outsourced the charger, or outsourced those other pieces, we couldn’t innovate as quickly. We couldn’t roll out things anywhere near as fast.

You made an early decision to switch from analog to digital controllers for the electric motor, which allows you to control the motor with software. How important was that decision?

Even we didn’t understand, in the early days, how much flexibility and agility that would give us.

There was a lot of hand–wringing, and it was a difficult decision to make the leap from the old pathway over to the new and really bet everything on it. But that decision set us up to put software in control of all of the key vehicle functions, and we are now unique in our ability to change those things remotely.

You’re referring to wireless updates. Last year two Model Ss caught fire after drivers ran over objects in the road. Tesla sent out a software patch that raised the height at which the cars travel on the highway. There have been no car fires since.

I’m totally convinced that the entire industry will go in this [wireless] direction. It’s only a matter of time.

You made an early decision to use small batteries, similar to the kind used in laptops, that cost less per kilowatt-hour of storage than the cells other automakers use. But whereas other automakers might have a few hundred battery cells in a pack, you have to use 10,000.

Other car companies think that large-format cells must be the way to go. But they’re more expensive and have worse performance. People think of thousands of cells and say, “I don’t know how to do that, and I don’t want to think about that.” It is a challenging problem. It’s harder to engineer a system to do it, but engineering is a one-time difficulty.

We started with a commodity [laptop] cell because we had 50 people and we couldn’t do anything else. But that plan has evolved. We’re now to the point that we’re working extremely closely with the cell manufacturers in designing customized cells with customized chemistry for cars.

You’ve used lower-cost cells, but because you chose to give your cars a 250-mile range, compared with less than 100 miles for most of your competitors’ cars, your cars are still expensive—people are paying $70,000 to over $100,000 per car.

People think the battery accounts for most of the cost of an electric car, but that’s not the case at all. For the Roadster [Tesla’s first car], the battery was already down below half the cost. Now we’re down to a quarter of the cost in most cases. We’re on track to getting to costs that will allow us to make a $35,000 car [the cost of the GM Volt] with a greater-than-200-mile range. It doesn’t require some mythical invention. All the pieces are fundamentally there.