Last year, we introduced the DRIVE PX self-driving car computer. And I said

three things. I said that, in the future, when you build cars, it's going to be a

lot more like a computer, that the ability for cars companies and system

OEMs to configure cars by connecting sensors and controllers one at a time

and evolve it over time is going to be more and more difficult, that you have to

approach it from a computing-system perspective. Software is going to be a

very big part of what you do, and ongoing improvements of the software is going to

be very important. The second thing I said was that computer vision technology

that had evolved up to this point using hand-coded features and engineers

crafting pieces of code to detect objects in the world are going

to be difficult to continue to advance to the point where we could have cars

that drives itself, and that we would use a technology called Deep Learning.

The third thing I said was that all of this computation capability is going to have to

be done in real time, because your car is driving through the streets. A lot of

things are happening around the world. And you want to be able to detect it,

understand what to do, and have the car keep you out of harm's way and navigate

safely. So the third thing I said was we're going to need supercomputing technology. All

of this we've been working on the last year. And today, I'm going to give you an

update on that. Several thousand man years of effort have gone into advancing

the state-of-the-art of self-driving cars at NVIDIA this last year. So first of

all, I'm going to show you -- I'm going to tell you the punch line.

Ladies and gentlemen, the NVIDIA DRIVE PX2. This is the world's first in-car

AI supercomputer, and it's designed to make it possible for us to realize the vision

of self-driving cars. It has 12 CPU cores,

four chips, each with Pascal GPUs in it with a combined processing capability of

eight teraflops, with a special new instruction design -- new set of

instructions designed for deep learning that makes it possible for us to achieve

24 deep learning tera ops per second. It's built with 16 nanometer FinFET.

All together, this entire supercomputer fits in your trunk very nicely, the size of a

lunchbox, 250 watts, all in that little tiny device.

The computational capability of PX2 is equivalent to essentially 150

MacBook Pros. Imagine 150 MacBook Pros of processing

capability in your trunk, all of this within the size of a school lunchbox,

250 watts. It needs to be water cooled. And the reason why it needs to be water cooled

is because we want to be able to make it possible for you to have this operate in all

kinds of severe conditions that a car could be enjoyed. Okay. So NVIDIA DRIVE PX2.

But let me back up a little bit and tell you the reason why we're doing this.

Our vision is to make it possible for us to finally realize the self-driving car.

Now, there's a lot of ways you can think about the self-driving car. From a

technology perspective, it's utterly amazing. If you ever have

enjoyed the process of having a car drive you to work, it's just

a really wonderful experience. However, technology aside, the contributions to

society is arguably incredible. Transportation, mobility is one of the

most important things in society.

It is central to how almost any culture in society is made. Humans are the least

reliable part of the car. We represent

almost all of the fatalities that are caused around the world, over a million deaths

each year. And the human is the most unreliable part of it. So whether we're

augmenting the human so that we could help people drive better, or replacing

the human altogether, self-driving car technology is surely going to make a great

contribution to society.

The second thing of course, if you can [have a car] that drives itself, all of a sudden, it's

possible for us to accelerate this entire movement of personal mobility as a

service. Fewer cars on the road -- the cars that are on the road are utilized more

frequently. The convenience of having mobility as a service is incredible. And

then, third, fewer cars -- the cars that are utilized more frequently will ultimately result

in having fewer cars in urbans, and we'll be able to redesign our urban environment and

make the neighborhoods more beautiful, more enjoyable. More [parks] will

be replaced by -- parks will replace parking lots. And the convenient factor

is incredible. So whether you think about self-driving cars from the money

that it's going to save, from the reductions in accidents, or the time

that it's going to save us, or just the incredible freedom it's going to provide

people who can't drive today for whatever reason,

self-driving cars is going to revolutionize society. Our vision is to create the

computing platform by which the entire automotive industry can realize this

vision