The Car is the AI computer

Massive change in car industry will produce high returns and a bubble

AI or machine learning is wide spread today in Silicon Valley. Until recently we were scratching our head on what that actually meant, who will profit from it and how money will be made.

Recently the bulb finally went on. It’s the car stupid!

AI has been around for a while. Google searches, Facebook face recognitions or Siri and Alexa are with us today. But they don’t matter. They’re nice to haves. The real thing is happening with cars.

Why is Padmasree Warrior working on a startup to build a self driving electric car. The former CTO of Cisco could get any sugar coated job in Silicon Valley. Why autopilot and electric car?

The women has worked at Motorola and then Cisco. Both companies have changed industries through computing. Motorola changed phones and Cisco changed Networking and communications. In both cases the change happened through silicon and software. Warrior is convinced that the same will happen with the car.

What exactly is happening right now?

First, there is Tesla. The company is successfully producing electric cars with advanced autopilot features.

Then there is Nvidia. Their processors are at he hart of AI computing. Nvidia is turning itself into an AI company due to their advantage in parallel processing.

As CEO Jensen Huang said recently; “we are running our chips for traditional work loads but we can use them for new types of workloads which happen to be useful in the AI space.” “We have a day job !” What he means is that Nvidia is in the position to do R&D for future applications since the R&D can be monetized today for traditional workloads.

Tesla and Nvidia are driving the sector forward.

The rest of the car companies are way back. This is a typical case of disruption, where existing car companies are not able to focus on the new thing since they have an old business to run.

No matter what happens to them, if Tesla keeps coming up with better and better cars whose quality is driven by software solutions the traditional cars will look like lemons. Something has to give.

You can buy a Model S with a performance of a Ferrari for less than a 7 Series BMW. And the car keeps getting better since software updates drive improvements.

The marketplace will not tolerate such a discrepancy for long. The Iphone was a similar shock to traditional phone makers and the car market is going the same way.

What’s interesting here is that AI can be used very effectively in cars. This means that AI will advance dramatically through the application of a car. The car is a clearly defined problem and it can be solved. It has a market and it will be felt my millions of customers. This is real impact.

But for this to happen we need a lot more than just algorithms and neural networks. It requires a whole value chain. The cars have to be made. Tools have to be made. New robots have to be made to produce the new cars. Components and semiconductors have to be developed to serve the new needs.

Examples of impact on businesses

For example, Analog Devices is in the Radio Frequency business. They serve markets such as defense or communications. But now there is a new problem. Cars want to see. And in order for them to see they need radar or liadar. Both are new types of problems and they will quickly become standard problems for car manufacturers. ADI or someone else will solve the problem for them. ADI recently bought a LIDAR business and will play an increasing role in this market.

Another example is Ansyss. It’s a company that does simulation technology. They sell simulation software to industrial clients such as car manufacturers. The new type of car company will have to do many more iterations and have a much faster model cycle. For this they need simulation technology. This is a a good example of what it means to introduce change in the value chain. Ansyss or others will solve the rapid iteration problem.

Infineon is an established chip company. They have decades of experience selling Integrated Circuits to car manufacturers and electricity clients. They are in a perfect storm since electricity and car manufacturing are converging. The first step in the new value chain will be a reordering of the semiconductor space. Suppliers will either adapt and ride the wave or become irrelevant. Since the problems will be new, new technology has to be developed. It’s interesting to see how established players will adapt.

Here is what’s going on in an analogy

Andy Grove in his seminal book “Only the Paranoid Survive” explains how inflections points in a business are detected. They are typically detected in middle or lower management. CEOs and their lieutenants are typically not the first to see it. That’s why they still talk about the same old stuff while their middle managers already sense the future. It’s interesting how this theme is playing out right now.

Take ADI or Infineon. Their sales are through the roof. But their top management talks about strategic decisions, good execution, strong markets etc. They are not saying what is really going on which is “the industrial value chain in cars is changing and we are profiting from this change.” The longer they wait acknowledging the real reason for their success the higher the risk they miss the wave.

Here is an illustration.

Let’s assume people realize that Beer actually cures cancer. Now let’s assume small groups of early adopters immediately start drinking more beer. Beer sales go up and established beer companies have good earnings. Typically management will tap each other on the shoulder and talk about execution and good strategic choices made in the past. They will neglect the cancer theme and talk about marketing, brand positioning and other typically beer jargon. Internally they will have meetings with middle management and listen to them saying how people are drinking more beer because it cures cancer. They will dismiss them and some of them will even leave out of frustration.

Next year the sales even accelerate and top management hires consultants to figure out what is going on. The consultants come back and confirm “It’s the cancer, stupid!”.

Now, top management changes the PR and talks about how well they are positioned to ride on this wave. But they still don’t trust the whole thing.

In year three sales more than double and the whole world starts drinking more beer because it cures cancer. Top management organizes an investor day in which they officially change their corporate message from “We are about Beer” to “We are about curing cancer”. From now on the company is the curing cancer business.

The whole investor day passes and everybody talks about cancer. The word beer doesn’t even come up. Old “beer concepts”, like marketing, advertising and brand positioning are irrelevant. The new new thing is “We cure cancer”.

This is exactly what is happening with Nvidia right now. The company was in the business of enabling graphics for computers. Because the Graphics Processors (GPU) is also good for machine learning, Nvidia has become the first choice when deploying a machine learning infrastructure. At their most recent investor day in San Jose in May 2017 there was only talk about AI and machine learning. The word graphics didn’t even come up.

We strongly believe that AI will be implemented in the car. The car is the killer app. With that other things will happen. Bur the car is perfect for AI as we know it. We can solve vision by applying deep learning and making sure the car does what a human driver would do. For this everything about the car has to change.

The Stack

  1. Sensors have to be developed so the car can see through cameras, lidars, radars and sonar.
  2. Mechanical innovations are needed to make sure the computer instructions get translated to the mechanical world with precision and very little latency.
  3. Manufacturing has to adapt to the new electronics. More precision, less waste and more focus on nanoscale problems.
  4. Manufacturing of cars will become more software like in the sense that interactions and error correction will happen more often and faster. Simulation technology has to help solve this process.
  5. Cars need computing power on the car and in the cloud. Connectivity has the be established.
  6. Car brands need to evolve. Performance will be determined by electric factors. For example torc is much stronger with electric cars.
  7. Cars will be sold similar to computers. Questions on the sales floor will be more like “How much memory?” How fast ist the processor?” “How much power does the IT in the car need?”

Implications for stocks

When inflection points like this happen stock prices move. It’s conceivable that a bubble will occur. Nvidia for example is worth something between 140 and 180$ a share. If a bubble occurs it will go as high as a 900 or a 1000 so it can fall back to 150 when the crash occurs. Typically after a bubble stocks drop by 80–90%.

We believe that electric cars and autopilot will transform the car market and introduce AI to real world.

We will watch Nvidia to determine where in bubble cycle we stand. Another candidate is Tesla.

Tesla is worth 450$ a share. For the bubble to happen and then burst Tesla has to go up to 3000$ and then crash back to 400.

These are extreme numbers. But the transformation in AI and cars is extreme and typically when that happens markets go to extreme levels.