They Don't Want to Be Nokia: Carmakers Built on Outsourcing Are Now Building Self-Driving Cars Themselves

Carmakers haven't spent this heavily on new technology in 20 or 30 years.

On September 12, an industrial park west of Detroit got busy. Ten white Ford Fusions — sold in China as the Mondeo — shuttled back and forth along the roads inside it.

A Ford employee sat in the driver's seat of every car, but all they did was sit there, watching the steering wheel in front of them turn on its own and chatting with the passengers behind them. After about a kilometer, the car came back to the starting point, picked up new passengers, and went around again.

It was the first ride Ford has opened to outsiders since it unveiled the Fusion self-driving prototype in 2014. This week, more than a hundred journalists and analysts from around the world, along with industry guests Ford invited, rode in those white Fusions.

In a talk that morning, Ford CEO Mark Fields announced the company's timetable: by 2025, a self-driving Ford will be in customers' hands.

The test Fusions looked much like the Mondeo you pass on the street, except for the four cylinders on the roof. They are lidar units made by Velodyne. As the car moves, each cylinder fires 32 laser beams in every direction and works out what is around it from how long the light takes to bounce back.

Raj Nair, who runs Ford's self-driving car development, calls lidar one of the technologies a car needs to drive itself safely. Google's self-driving cars use the same company's hardware.

To lock that technology down, Ford bought a stake in Velodyne. Last month the lidar maker took its first outside investment since it was founded — US$150 million, led by Ford and Baidu.

Ford has not said what else is in these cars. Apart from the lidar, most of the core parts sit in the trunk.

You don't have to open the trunk to know what Ford will be using in 2025, if the plan holds. In less than six months, Ford has disclosed five investments tied to self-driving technology.

• May: Pivotal Software, which builds supporting software for in-car infotainment systems.
• July: Civil Maps, HD maps made for cars to read.
• August: besides Velodyne, Ford also invested in SAIPS, which works on machine-learning algorithms and image analysis.
• September: Chariot, a commuter carpool service.

When a technology company moves into a new industry, standing up a semi-independent team and buying new technology in bulk is routine.

Ford, and the car industry as a whole, has not done it in 20 or 30 years. Carmakers spent those decades doing the opposite: spinning off parts development and manufacturing and outsourcing them. Ford investment news used to mean Ford buying another car brand, like Jaguar and Aston Martin in the 1990s.

For decades, carmakers handed their engineering to someone else

"We've vertically integrated — we're more like the old Ford than Ford is." That is how Tesla CEO Elon Musk, in an earlier interview, described what he had been doing.

Tesla today designs its own battery management system, sets its own battery specifications, collects its own map data, and has hired hardware engineers to build a chip for its self-driving system.

It really does look like the Ford of old. In the 1920s the Model T made Ford the most influential carmaker in America, and its grip on how cars got built ran deeper than Tesla's does now.

Besides the assembly plant at Highland Park, Ford owned a dense line of steel mills, engine plants, glass works and tire factories along the River Rouge in western Detroit — 127 buildings in all. Every one of them belonged to Ford.

Henry Ford even told suppliers exactly how to build the wooden crates that carried parts to him. Once unloaded, the crates were broken into slats that went into the car body. The scrap and sawdust left over became Kingsford charcoal — another company the elder Ford founded, and one that is still alive today.

Today, never mind the shipping crates: on many models Ford does not fully own even the technology that decides how the car drives. The brakes on the F-150 Raptor, its best-selling pickup in the US, are bought from Bosch.

A supplier that owns the patents can usually widen its margin in a deal with a carmaker. More and more of the industry's new technology is born inside big suppliers like Bosch, Delphi and ZF.

In 2015 Bosch's sales reached €70.6 billion, roughly 30% of the annual revenue of Toyota, the largest carmaker in the world, and more than that of Renault-Nissan, which ranked fourth in global vehicle sales that year.

American carmakers went from owning steel mills in the 1930s to not fully controlling their own powertrains, and they did it to cut costs.

Toyota had been raising the share of parts it bought in ever since the end of World War II. Founder Kiichiro Toyoda worked out a system for managing outside suppliers, treating standard parts and core components differently.

The Toyota Camry, a US bestseller in 2007, had several hundred parts suppliers: airbags from Meian, anti-lock brakes from Bosch, pistons from Mahle.

Starting in the 1990s, the American industry copied the arrangement wholesale. In 1999 GM spun off its parts arm, Delphi. "Making everything yourself is no longer an advantage," GM chairman John F. Smith Jr. said at the time. "We need to become a faster company, focused on the core business of building cars." A year later Ford cut loose Visteon, its own parts operation.

If carmakers gave up powertrain technology gradually, they never took in-car infotainment seriously in the first place. Ford built its SYNC system with Microsoft, then switched partners to BlackBerry's QNX. Mostly Ford wrote the requirements and let software companies like Microsoft and BlackBerry build the thing.

With Apple CarPlay and Google Android Auto in every new car, even the interface an owner looks at all day is Apple's and Google's call.

It isn't only Ford: Europe's and America's big carmakers are doing self-driving in-house

Carmakers buying into ride-hailing apps and maps does not mean they have stopped chasing savings. Ford also said this week it would move small-car production out of the US to Mexico.

Today, apart from a handful of companies backing hydrogen and batteries, most of the industry's technology money goes to self-driving.

At CES this past January, HERE Maps put up a tent of several hundred square meters outdoors. HERE began as Nokia's digital mapping business, providing navigation to Nokia phone owners from 2009 on.

In August 2015, Nokia — long out of the phone business — sold the maps to three luxury brands: BMW, Mercedes-Benz and Audi. But they did not pay €2.8 billion to push in-car maps on drivers who had long since settled into navigation apps from Google, Apple and Amap.

Most of the screens in HERE's CES booth ran demo animations about self-driving. What the company sells now, in the words of the staff there, is a map for cars to read.

On the map you use, the thinnest line might be an alley. A map built for cars has to know whether the line marking a lane is dashed or solid, and to recognize a sign that says "bus lane" or "restricted hours."

A self-driving car needs more than a map. It needs lidar to see the road around it and AI to decide how to drive. Beyond Ford, GM, BMW, Mercedes-Benz and Volkswagen have all set their sights on these technologies.

The latest numbers from CB Insights, which tracks investment and funding, show GM, BMW and Mercedes-Benz as the most active carmaker investors through July this year.

Investing aside, Ford, GM and others are building semi-independent teams inside the company to work out how to make self-driving happen. Between 2010 and 2014, R&D spending at the mainstream carmakers rose 67%, to US$137 billion.

The hurry comes from a fear of becoming the next phone industry

It isn't hard to see why a car company would rather give up powertrain technology than lose the race for self-driving.

For more than a century the industry has sold cars on words like comfort, performance and looks. But if the car drives itself, there is no pleasure of driving left in taking one home, and buying a self-driving car starts to look a lot like hailing an Uber or a Didi.

Uber sees it the same way. CEO Travis Kalanick once said drivers were his biggest cost, and that if Tesla could build self-driving cars he would buy every one of them.

Before long he changed his mind and decided to own the technology himself. Uber first poached 40% of the researchers at Carnegie Mellon University's autonomous driving lab, then struck a deal with Microsoft to take its Bing Maps engineers.

If Uber owns self-driving technology, it can pick whichever carmaker offers the right price and have the cars mass-produced. The new cars would not be sold to consumers; they would drive straight out and start earning.

Once self-driving technology matures, the business model for cars may change. No more pouring money into advertising in the hope that a customer spends 100,000 or 200,000 yuan in one go.

Plenty of people will still buy cars, but a good share of vehicles will become a service you summon. And the company that builds a self-driving car no longer has to worry about whether it sells — the car earns money from the day it is finished.

If carmakers do not own the technology by then, they will have to lean on technology companies, and car manufacturing ends up as a slightly more sophisticated version of FIH Mobile, the contractor that assembles other companies' phones.

The phone business is the warning. Apple and Google rebuilt the smartphone around their own operating systems. A few years later Nokia, which made its own system, sold off its phone business, and so did Motorola, which had thrown in with Google.

China has no shortage of phone makers, but 90% of the hardware profit in the industry belongs to one company, Apple. And free as Android is, using it outside China means Google takes the cut from the games, apps and advertising on the phone.

That is probably also why every mainstream carmaker is investing in self-driving, and why Toyota, BMW, Volkswagen and GM have each put money into car-sharing. As long as a person is still driving, carmakers cannot compete with Uber and Didi, which have signed up tens of millions of drivers apiece. If nobody has to drive, the carmakers' odds in services improve.

Technology companies are fighting carmakers for the same future. Tesla's semi-autonomous Autopilot got on the road first. Google has been testing since 2009, and Apple recently redirected its car project toward self-driving systems.

Their problems are not small either. Tesla still has not cracked mass production. Google's and Apple's car projects are years from arriving, and their engineering teams are already in turmoil. A car really is a far more complicated product than a phone.

The carmakers are not in much better shape. Next to a technology company, the old carmaker-and-supplier arrangement is slow: an all-new model normally takes five to eight years. Eight years was enough for the iPhone to carry Apple from nothing to the most valuable company in the world and put smartphones in the hands of billions.

Investing in companies and building teams means changing that pace and figuring out self-driving inside the company. For an industry that cannot get in-car entertainment right, that will not be easy either.

"The driverless car will have as much impact on society as the assembly line Ford invented a hundred years ago." A month ago, Ford CEO Mark Fields said as much into CNBC's cameras.

The question now is who gets to be this century's Ford first.

Cover image from Carkeys