As the Apple Car dies, Tesla pushes the envelope on self-driving

I have to admit that for Wednesday's much-anticipated Tesla announcement, I was hoping for a new vehicle - the SUV version of the Model 3 mass-market car, the so-called "Model Y."

That didn't happen, but we got something arguably better: a sophisticated new suite of self-driving hardware technologies for all forthcoming Tesla vehicles, plus what CEO Elon Musk described as a separate "supercomputer" brain to coordinate all the cameras and sensors.

Together, these will make full autonomy a reality at some point in the future, Tesla said.

It won't come cheap: the complete self-driving package will cost $8,000. But if you think about it, that's astonishingly inexpensive to deliver the car of the future and be the first to cross the finish line on truly full autonomy.

Tesla's announcement went beyond Apple doubling down on it Autopilot self-driving tech after a fatal accident in May - Musk and the team are trying to make good on a pledge he made last year, on Twitter, to hire "hardcore" software engineers to work on Autopilot advancement, engineers who would report directly to him.

To get from where we are now - Level 1 or 2 self-driving - to Level 4 or 5 full-autonomy will require some breakthroughs, but it now looks as if Tesla is gearing up for just that.

Meanwhile, elsewhere in Silicon Valley, it appears as thought the mysterious Apple Car "Project Titan" has flamed out amid incoherent management and an inability to figure out whether to build an entire car or to focus on autonomous driving and a disruptive in-vehicle user interface: the much-discussed iPhone on wheels.

A tale of two cars

The contrast is instructive. At the moment, we're seeing a second great proliferation of automotive startups, many proposing to create self-driving electric vehicles. None are building or selling any cars, however. Google's self-driving cars have racked up millions of miles on actual roads, but nobody at Google seems to know how to commercialize this impressive technology.

Tesla, however, will probably sell around 80,000 vehicles this year. By 2018, it plans to sell 500,000.

From here on, all those new Tesla's will possess the hardware and software to be fully self-driving. Critically, those technologies will be completely integrated with Tesla vehicles, through the manufacturing process. That's important because a bolt-on self-driving technology might work with one type of vehicle but fail utterly on another.

The bottom line is that if you want to go all-in in autonomy, you have to be a carmaker, with all the traditional manufacturing capability that such a business identity requires.

Some new entrants to the space probably think they can dodge this, avoiding the costly process of constructing their own factories and getting good at assembling conventional automobiles - or at least conventional looking automobiles, machines with doors and tires and windows.

But as Tesla has now demonstrated, that can't be done. Creating a fully self-driving car and getting it on the road means that the "car" part has to come first.