In 2012, Kentley-Klay stumbled on a blog post about Google’s self-driving car project, then pretty much the only one in the field. He saw the company’s prototypes as unsightly half-measures.. Then, one day, he walked into his Melbourne office (animation, video production) and announced he was off to America to fulfill his driverless dreams.
In a move that some will call devious and others will call ingenious, Kentley-Klay reached out to some of the biggest names in the field and told them he was making a documentary on the rise of self-driving cars. The plan was to mine these people for information and feel out potential partners. His first “interviewee” was Sterling Anderson, then a robotics researcher at MIT and later Tesla Inc.’s self-driving car chief. “I played the oldest trick in the director’s book: the vanity card,” Kentley-Klay says. “I showed up at MIT with a Canon and a bullshit microphone and interviewed Sterling for two hours in a grassy field..."
> Wanting to preview his line of shoes at Market Week at the New York Hilton, but unable to afford the purchase of a hotel room or showroom to display his items, Kenneth Cole inquired about parking a trailer two blocks from the Hilton Hotel. Upon discovering that permits for trailers were only granted to utility and production companies, Cole changed the name of his company from Kenneth Cole Incorporated to Kenneth Cole Productions, and applied for a permit to film the full-length film, "The Birth of a Shoe Company". In two and a half days, Kenneth Cole Productions sold forty thousand pairs of shoes, while chronicling the beginning of the company on film. 
Which is why it's still called Kenneth Cole "Productions" to this day.
If I pay you 100k then you spend 50k buying land that’s not a “loss”. When a company spends 50k on R&D that’s generally an investment not an operating expense. Plenty of other dodges exist for physical assets via depreciation games etc. Even just buying more brand averting is an option to increase capital without being taxed on it’s creation.
For a business that really doesn't make a difference. If you have $1 billion in revenue but spent $999,999,999 in order to generate the revenue then your business income is $1.
Or the complete logical extreme. If I sell a dollar bill for $1 that's not income that should be taxed.
You always have overhead so you can’t actually sell 1$ for 1$. Breaking even requires selling assets for more than their worth to cover transaction fees. We let companies deduct those fees, but what’s a fee and what’s an investment gets blurred. Especially when most companies don’t suddenly get destroyed after a single transaction. The classic razor and blade model being a prime example for hiding profits.
Sure, cash flow may make it seem minimally profitable, but that’s only relevant in this quarter or if you run out of cash. Spend X million on software development or brand advertising and can have a tax free transfer from income to capital. Leverage that for a few decades and suddenly your business is worth 10x as much without any apparent profit. Clearly an asset was being invested in.
I gave up on the first four...
High school lunch-table politics don't end at 18, unfortunately.
For one, I'm glad to live in a world where those kind of things still exist.
Lying your way to success?
I'd never look at the Odyssey and mistake it as something to emulate. That people do is kind of surprising.
But here is an interesting take. Amazon is an formidable acquirer and probably the only big-tech that knows how to squeeze from their M&A activity.
They probably saw this years ahead. Many companies were going to develop technology in this space and it was likely better to acquire one of those than to develop in-house. And the reason is very simple: Amazon doesn’t invest in greenfield projects that are not customer facing. Look it up. Almost all the behind the scenes impressive tech they have was acquired at certain time.
For example. AWS proprietary chips come from their acquisition of Annapurna Labs.
All their warehouse robotics come from their acquisition of Kiva Systems.
All their modern video tech comes from their acquisition of Elemental.
Their customer facing acquisitions also play a similar playbook but in a more -fill the market gap- kind of way. “We need a supermarket, let’s buy Whole Foods”.
“We need routers for the Alexa ecosystem, let’s buy Eero”.
Also kind of interesting that they are an opportunistic buyer (almost vulturous) that most of the times only pulls the trigger when the acquisition target is in their lows.
I don't think this is very true, they are just very good at turning around and selling the internal tech.
The underlying technology for AWS was developed internally and it was not customer facing at first.
Similarly if you look at their internal data platforms, they have invested heavily even though it's not customer facing (yet, because eventually they will start selling services inspired by the internal systems via AWS).
I remember watching a presentation on how they invested large amounts of money into prototyping FC robotics technologies.
If you look at AWS Kinesis or EventBridge, the technologies were developed for internal usage before they were sold. There is also considerable investment into distributed consensus technologies that hasn't been turned directly into products (yet).
You're kind of right in that Amazon doesn't spend much time working on greenfield projects that don't have any forseeable business value (in contrast to say Google), but that is in many way a product of their organizational structure where two-pizza teams are supposed to do 'research' as part of their product work rather than having a pure R&D division.
Will this myth ever die?
Plus, AWS no longer use such narrative anyway.
Edit: that’s not the right podcast. This was before he went to AWS. I can’t for the life of me find the interview.
But if someone with deeper insight than someone who has been working at Amazon/AWS for -3 days (I start Monday), I’m glad to be proven wrong.
So, even without a mandate, the market forces at play in terms of proposing roadmap projects and choosing what to invest in, will naturally tend to favor those solutions with can pull double-duty (solve an immediate problem/pain-point that Amazon already has, and with a clear path/potential to be converted into or evolve into an external sell-able service too).
Unsurprising since Bezos himself worked in a Hedge fund. The Autonomous Driving hype train is slowing down, so this is the perfect opportunity for this move.
Interestingly, with this move, Bezos can soon rival Musk for the next 'billionaire super genius rock star' position.
I've been quite dismissive of Blue Origin in the past, their pace of development seemed so sluggish compared to SpaceX, but I think that was somewhat unfair. It seems Bezos only really started pouring money into the project a few years ago though. BE-34 does seem to be behind schedule, or at least the launch vehicles are by a year or two, but that's par for the course for big rockets.
He funds blue origin with more than a billion $ every year from the sales of his own amazon stock. Without amazon filling his pockets he can't finance blue origin.
He puts more of his money in blue origin every year than the amount amazon will pay for zoox
Zoox in particular was a LARP of an autonomous driving company. Until any of these companies start manufacturing, they’re literally LARPing. At least Tesla shipped an actual product.
Anyhow, I haven't seen the trend of calling out an entity as being a LARPer. But I have to admit, it's kind of funny. It does make me giggle :)
Not really sure how you're using "squeeze" here, but Instagram and Whatsapp were both, with the benefit of hindsight, phenomenally successful acquisitions.
>Amazon to acquire Yap, move into speech recognition?
>Amazon has reportedly acquired Evi for voice-guided search
I think Annapurna was more strategic than you think.
* James Hamilton in 2010: https://perspectives.mvdirona.com/2010/10/datacenter-network...
* Annapurna acquired in 2015.
* In 2017 they reveal that every EC2 server has a Nitro (Annapurna) chip for networking: http://www.brendangregg.com/blog/2017-11-29/aws-ec2-virtuali...
Chips are peculiar because they matter a lot in a customer facing way but they are not a 100% transparent feature. I think AWS customers couldn’t care less what their instances are powered by as long as they are 1) Fast 2) Cheap 3) Secure. But the chip happens to influence those things heavily, so they become almost like referential values to aggregate those characteristics in a way that’s easy to market.
Whenever they take their cars out for a test drive, they sit in the middle of the road in the parking lots. If you aren't going to be driving around anytime soon, just pull into a parking spot, instead of blocking all the empty spots and making it more difficult for people to drive around you.
I've been on booth duty and sales duty before and you should never snub people like this - ever. It is hurtful and people remember it, I remember it still so many years back. There are always more graceful ways of doing things.
nit: You should edit this and change Zoom to Zoox before the edit window closes.
Suffice it to say that I don't think self-driving car tech is ready for city streets. Zoox never put anyone in harms way, but damn did it hold up traffic if there was an "out of place" person.
Self-driving tech may be close to cruise control 2.0 - that is highway exit to highway exit navigation - but I have yet to see a self driving car handle a city street with actual people on it.
This is why the main first application of autonomous driving is going to be long haul trucking.
The current regulations about duty time for humans in long haul trucking create inefficiencies. Now imagine that there are facilities just off of highway on- and off-ramps where truck drivers hang out. The autonomous vehicle pulls off the highway into a lot. A human then gets in and guides the truck to it's final destination.
Truck drivers get an immediate lifestyle increase (no more multiple days on the road, having to sleep in the truck, etc), labor costs drop dramatically, and travel times for cross country trips also drop dramatically.
The driver market for CDL truck drivers will turn into the "harbor pilot" model they have for large ships.
Even so, these "lifestyle increase" jobs would only exist until autonomous driving gets better. Even if that's two decades it's not something you'd want to rely on as an occupation.
Not saying it's the wrong way to go - it absolutely is the right way to move logistics forward - but there will be significant impacts in a country where "Truck driver" is one of the top jobs in just about every state that can't be explained away with "lifestyle improvement"
Disclaimer: I am waiting for my CDL Class-A test ;)
In between, gas stations will have full service again for diesel. There will be some way to automatically charge the truck for fuel.
Early employees who were dreaming of a windfall aren't going to get one I guess, and anyone who left the company or was fired and exercised is totally screwed.
The typical result of M&As is, by and large, that the workers lose.
That's not what's happening here, though - Zoox's assets are worthless without their employees, so its in Amazon's interest to not screw them over completely.
These guys are supposedly launching their ride hailing fleet this year on public roads.
"unorthodox entrepreneurial zeal of Kentley-Klay, an Australian native with no prior automotive experience." was the key mover on all this and "touted Zoox’s strategy of building its own vehicles for full autonomy as wiser than the standard approach of retrofitting existing cars that Alphabet Inc.’s Waymo and others are taking."
Their CEO did attack silicon valley for choosing the path of fear "at the expense of profound progress for the Universe."
Heady stuff! Will be very interesting to see their launch if all this talk pans out! Full new vehicle assembly from scratch PLUS autonomy! Very impressive.
I mean, this a bad prognostic sign if there ever was one in an engineering company.
Whether or not Zoox is a bust from the investors perspective, the synergy with Amazon seems obvious.
Also, what happens to employees? Presumably, their stock is worthless. Will Amazon just give them all new stock packages to keep them on?
Likely they didn't have a business model and so were a fire sale.
Amazon will almost assuredly eye this for autonomous delivery. They have the cash, the legit need, the ambition ability to make it work. So it can be commercialised 'internally'.
Unlike Google who have to build a business around their self-driving cars.
This is one of those 'tech may eat the world' things that should legit scare FedEx because I doubt FedEx will be able to compete on terms like this.
> Zoox planned from the beginning to develop the technology, build a car and operate a robotaxi service. Not even Waymo is trying to bite off that much (although Cruise is).
Is Tesla intentionally omitted or am I missing something?
I think the tech press doesn't believe that Tesla is building a self-driving product.
I don't know if camera based computer vision by itself will ever be enough for full self driving. I think we can demonstrate that it is not enough at this time, and we can also demonstrate that human vision also fails at times; human vision has the benefit of some extremely specialized "software"
Rebuttal: If you had a 3D camera sitting in the driver's seat of a car and hooked up the control to a remote controller, I believe that car could be drive-able. A sufficiently advanced perception/control system could theoretically use the same 3D camera feed and replace the driver. Not saying it's easy, just that all the necessary info to drive a car safely can be contained in camera feeds.
There are many tech and non-tech press which didn't believe in anything posted on the "master plans" published by Tesla, and they are still in denial.
Meanwhile, Tesla has been making huge leaps of progress and collecting billions of miles worth of data to improve their systems. What is available right now, and has been so for a couple of years, is unmatched by any of their competitors.
I am interested to see how Zoox and others will compete against massive amounts of data on corner case scenarios and real-world driving.
The data from real world is far more important than simulations.
Take a look at how long it takes to deliver a package today. The driver drives up, parks. Gets out, goes to the back and finds your package. Grabs it, walks over to your door, drops it off (and takes a pic), then walks back, closes the door, straps on the seat belt, looks up directions to the next stop, and then drives off.
With an automated truck, you wouldn't have a driver; you'd just have a delivery person. While the truck is driving, the delivery person (DP) would grab the next package. As soon as the truck stops, s/he hops out with the package and drops it off. Then, the moment s/he climbs back on and and sits down, the truck is off to the next stop.
By my estimate, it would cut down the delivery time to half of what it is currently. IOW, a DP + truck could deliver 2x as many packages. You can expect the tech to pay for itself in just a few years.
Curious to see how that pans out.
His previous experience was running an animation studio in Australia...so there's that too...
Further, I would think a firm promising enough to seed a self driving program to power Amazon’s delivery fleet would be worth more than a billion dollars.
Would love to hear some thinking on this.
Update: There’s a comment up thread to the effect that buying seed a growing their own program to support their internal operations is standard operating procedure for Amazon that provides a number of good examples.
General public expects driverless cars, car companies want level 3 automation, Google says there is only level 2 and 4 which is probably right, and the technology is going to stay level 2 for at least better part of next decade if not couple decades.
What’s working is Level 4 betas in Level 2 mode with driver monitoring for safety, what I would call a “driving simplification”.
“Simplification” applied to delivery vans offloads what used to require driver competence to machines thus improve consistency, and that will work for Amazon.
Level 2: driver always pays attention
Level 3: car pays attention, may request driver
Level 4: car do not require driver
This seems simple enough, but Level 3 requires car to be able to drive entirely by itself + be able to predict situations minutes into the future, including situations that cannot be predicted. That requirement far surpasses Level 4, or laws of physics known to humans, Level 3 classification is thus invalid and there are only Level 2 and 4.
How far it has to predict something into the future is however long it takes for a person to safely take over - let's say 10 seconds. The thing is though, a Level 4 system also has to be able to predict things into the future in terms of erratic drivers and pedestrians - so the problem is surprisingly similar.
If you want to know what's really going on with self-driving, talk to the prediction teams at any of the companies. It's pure research, because how can you predict the future? And it's not a hypothetical problem, because pedestrians do weird shit and you need to predict if they're going to walk in front of you 5 seconds from now all the time (at least for urban driving). Humans do this instinctively - we slow down in uncertain situations and are constantly monitoring it for changes. When the robots try to do this, they slow down and then just stop because they can't really guess what's going to happen next. For computers, I'd argue it's roughly equivalent to solving hard AI.
So they worded that that you may have either:
1) no wheels, or
2) wheels that may shear off, or
3) wheels that may shear off but continue to support weight for a while, or
4) wheels that don’t shear off under normal conditions, or
5) wheels that don’t shear off under any circumstances.
Simple enough right?
I don't drive much, but I still hit cases like this periodically, especially when driving through the canyons. From the user's perspective it's effectively the equivalent of taking transit, with very occasional non-realtime takeovers. By contrast, L4 can do all the things that don't require a human in the car at all, like cargo or getting to their next pickup.
If you're a research organization developing self driving cars, you will find yourself stuck at level 3, because level 4 is really really hard. Going from level 3 to level 4 means getting rid of any problem that would keep you at level 3.
Disclaimer: I work for GM, not on SDC, this is solely my opinion.
I definitely agree it won't be perfect, some people would indeed have a real bad time, but that still doesn't mean an improvement over the current status quo wouldn't be a seller. Perfect is the enemy of good, after all.
Current level 1 and 2 systems wouldn't have been nearly as popular as they are if consumers didn't want them. Honda wouldn't have made it standard on all their cars, etc.
Because investors are assumed to be sophisticated adults who can read and do their own research, or hire lawyers to do so on their behalf if they are incapable. Every investor, those who put in money and those who put in labor hours are assumed to have gotten the best deal available to them.
Think about this situation: a company is in dire straights, and will likely go bankrupt and be worthless. The founders and the employees still believe they can turn it around, however. As an investor, you may be unwilling to invest with a 1x liquidation preference, given the high likelihood you lose all your money. So the company and the investor may negotiate a 2x liquidation preference as a way to sweeten the deal and induce the investor to invest in a riskier than usual proposition.
Similarly, right after the dotcom bust when many investors lost their shirts, terms became much more investor friendly because the risk was perceived to be higher.
At the end of the day, the terms are negotiated between the company and the investors, and it’s a market - if the environment is founder friendly and there’s plenty of capital looking to invest (like today), it’s almost always 1x. If capital is scarce, you take what you can get, otherwise, no capital.
The founders are well aware as they negotiate these terms. The rank-and-file employees may not understand the company’s liquidation preference stack, but they can always ask, and if not given or satisfied with answers, ask for more cash comp than equity. It’s again a free market.
Not even close, Alibaba handles 200 million packages per day and is building out for one billion.
Amazon is barely a tenth of that current volume.
> "Zoox has already raised about $1 billion in funding — and once was valued at $3.2 billion"
More and more I'm reminded to search for context when presented with a big-looking number.