Feb 20, 2019

Axios Navigate

Joann Muller

Good morning! Thanks for reading. Please share this newsletter and tell your friends they can subscribe here. If you have tips or feedback, just reply to this email.

Today Expert Voices contributor Bibhrajit Halder looks at the race to develop a new generation of AI chips for autonomous vehicles.

1 big thing: Amazon's AV bet could make deliveries even cheaper

Illustration: Sarah Grillo/Axios

Amazon has been pouring money into AV technology lately, a strategy that could squeeze its retail competitors by allowing it to control the entire shopping process from click-to-buy to delivery, Axios' Erica Pandey and I write.

What's new: The online retail giant has been fairly quiet about its AV ambitions until now. But with a rapid-fire series of investments, Amazon is declaring its intention to automate shipping and logistics every step of the way — from its warehouses to highways to your doorstep.

The big picture: Shoppers have become accustomed to almost free, lightning fast delivery. Without a driver, AVs would be able to deliver goods even cheaper and faster, KPMG researchers say, increasing pressure on brick-and-mortar retailers and triggering even more online shopping.

  • McKinsey predicts autonomous deliveries will slash retailers’ shipping costs by 40%.

Context: Several big-name retailers are already experimenting with driverless delivery.

  • Kroger is partnering with the newly flush Nuro to deliver groceries with robot vehicles in Arizona.
  • Walmart has enlisted Waymo, Google's self-driving car unit, to bring shoppers to its stores in autonomous minivans, though it isn't delivering goods in AVs.
"The economics of pulling the driver out of the vehicle to get the last mile or the last 50 feet is astonishing. That's why everybody is investing in AV technology."
Gary Silberg, automotive sector leader, KPMG

Yes, but: Amazon is bigger and stronger. Its Prime membership business, worth an estimated $100 billion per Morgan Stanley, has already overwhelmed the competition with its speedier and cheaper delivery.

  • Last holiday season, the company kicked off a race to the bottom on shipping prices by offering free delivery for weeks.
  • Amazon already has thousands of trucks, a handful of airplanes and even some drones that it uses to deliver packages.
  • Amazon faces the same challenges as others — AV technology is new and widespread deployment is not a given. But adding its own fleet of autonomous trucks could drive its prices down even further and thwart rivals' attempts to leverage AV tech through Waymo, Aurora or others.

Details: Amazon formed a small team to investigate driverless technology several years ago and partnered with Toyota in early 2018 to explore AV deliveries. It has also been spotted using self-driving trucks to haul cargo in Arizona. But in recent weeks, they've doubled down...

  • In January, Amazon introduced Scout, a cooler-sized, electric robot for last-mile deliveries.
  • Earlier this month, it joined a $530 million investment round in AV startup Aurora Innovation, led by an all-star team of engineers from Google, Tesla and Uber.
  • A week later, it led a $700 million investment in Rivian, whose modular electric chassis can be adapted for virtually any type of vehicle.
  • With Aurora's self-driving system and Rivian's flexible "skateboard," Amazon could potentially fashion different sized AVs for any purpose.

The bottom line: Amazon has a history of expanding into areas within its own value chain wherever it can be successful, notes Roy Bahat of Bloomberg Beta. “Automated vehicle technology is now mature enough that it's within the striking distance of Amazon's long arm.”

2. Next-generation AI chips could speed up AV computing

Illustration: Rebecca Zisser/Axios

AVs have to interpret sensor data, determine their next moves, and follow through on them — all of which requires exceedingly complex AI. To meet those demands in real time, computing for AVs will have to happen onboard the vehicle, AV veteran Bibhrajit Halder writes for Axios Expert Voices.

Why it matters: The alternative to onboard computing for driving functions would be vehicles relying on unstable network bandwidth for cloud computing while cruising at highway speeds.

  • A specialized AI chip market has emerged to create platforms that can perform these complex computations almost instantaneously, while using as little power as possible.

Where it stands: More than 70 companies have entered the AI chip market in the last few years, and another 100 chip startups have been announced. The next generation of chips could have an array of uses, beyond advancing AV computing power.

  • Nvidia is dominant in the GPU space. It offers a computing platform with a plethora of chip offerings for deep learning that take advantage of GPUs' massive parallel computation abilities. Next-generation technology, however, will introduce more efficient chip architectures designed expressly for deep neural network computation rather than graphics processing.
  • Google has developed a circuit called a Tensor Processing Unit (TPU) that has run common neural networks 15 to 30 times faster than a comparable GPU, and used far less power to do so.
  • Startup Graphcore develops accelerators and software framework together, which arguably combine to make the fastest and most flexible platform for AI applications. Graphcore has developed an Intelligence Processing Unit (IPU), which offers promise for AV use.

What to watch: Roughly 12–15 companies are pulling ahead with their next-generation AI chips, but solidifying a lead will require producing and scaling them as well. This could favor established players or motivate smaller companies to merge with larger ones, as Nervana Systems and Intel did.

Go deeper: Read the full piece.

Halder is CEO of an early-stage AV startup and has worked on AVs at Ford, Caterpillar and Apple. He is also a member of GLG, a platform connecting businesses with industry experts.

3. Musk offers another AV prediction

Tesla CEO Elon Musk. Photo: Nora Tam/South China Morning Post via Getty Images

CEO Elon Musk yesterday said Tesla's full self-driving feature will be completed by the end of 2019.

My thought bubble: Don't count on it.

Why? Musk has made bold predictions before, telling Fortune in 2015 that Tesla cars would drive themselves within two years. So it's prudent to take his latest prediction, during a podcast interview with money management firm and Tesla investor ARK Invest, with a grain of salt.

Key quotes, from the podcast per CNBC:

“I think we will be ‘feature complete’ on full self-driving this year, meaning the car will be able to find you in a parking lot, pick you up, take you all the way to your destination without an intervention this year. I am certain of that. That is not a question mark.”
"However people sometimes will extrapolate that to mean now it works with 100 percent certainty, requires no observation, perfectly. This is not the case."
"My guess as to when we would think it is safe for somebody to essentially fall asleep and wake up at their destination? Probably towards the end of next year. That is when I think it would be safe enough for that."

Between the lines: Until then, of course, drivers need to stay engaged and be ready to take control of the wheel at any time. The problem is that some people have already fallen asleep behind the wheel of a Tesla, mistakenly thinking Autopilot will deliver them safely to their destination.

Meanwhile, Tesla rivals are tempering expectations for self-driving cars, acknowledging that the challenges are more difficult than expected.

4. Driving the conversation

Uber impact: Uber and the ongoing erasure of public life (Nikil Saval — The New Yorker)

  • Worthy of your time: This is a thoughtful (not knee-jerk) essay on how Uber has affected cities and its residents.
  • "The app’s interface — that empty map — declares its priorities: the individual, the vehicle, and a place to be. It erases public space and public lives. The public good is not far behind."

Bikes are out: Lime dropped the 'bike' from its name; now the bikes are disappearing (Melia Russell — San Francisco Chronicle)

  • The big picture: Scooters are a better business than bikes, it turns out. It must be true: Lime just raised another $310 million in a funding round that valued the company at $2.4 billion.

Hyperloop: A Real Tube Carrying Dreams of 600-M.P.H. Transit (Eric A. Taub — New York Times)

  • My thought bubble: Is it just me? I'm in no hurry to be hurtled in a vacuum tube at speeds that could exceed 600 miles per hour.
5. 1 uplifting thing

Elisha Graves Otis shows his first elevator in the Crystal Palace, New York City, 1853. Photo: Getty Images

For all the talk of how driverless vehicles will transform our cities, Forbes contributor Carlton Reid reminds us that one already has — the elevator.

The big picture: Elisha Graves Otis' invention of the elevator brake in 1853 ushered in the Skyscraper Age, which led to the rapid growth of cities in the 1870s and 1880s, according to Reid.

  • The first elevators had operators, but by the 1920s, they became "driverless" when electronic controls replaced them.

Why it matters: People were afraid of elevators just as they fear self-driving cars today. As Reid writes...

The widespread acceptance of vertical transportation inside buildings happened only once people became convinced they wouldn’t hurtle to a surefire death.
And Otis himself had led the convincing. High above the crowd at the New York World’s Fair in 1854, the 42-year-old industrialist had an assistant cut the only rope suspending the platform on which he was standing.
To gasps, the platform, and Otis, dropped – but it and he were stopped by a spring engaging a toggle on a cogged rail. The brake did its job, and the failsafe elevator had arrived.

The bottom line: There are lessons to be learned from the public's acceptance of elevators, Expert Voices contributor Stan Caldwell of Carnegie Mellon's Traffic21 Institute emails Axios. He says people don't have to wait for AVs to be perfected before experiencing the benefits of partial automation and adds...

"Elevators relied for decades on human operators, they still require a 'call' button for human intervention, and they cannot be used in a fire. Yet millions of people use this vertical automated vehicle daily while understanding its limitations."
Joann Muller