Hacker News new | past | comments | ask | show | jobs | submit login
One year old Pony.ai raises $112M Series A to build autonomous car future (techcrunch.com)
65 points by rbanffy on Jan 15, 2018 | hide | past | favorite | 28 comments

$112M for... geofenced self-driving? Is that the main differentiator, or did I read that wrong?

I don't get their competitive advantage. Sure, he worked at Baidu and led a team there, but this is a LOT of money for a very small chance at anything competitive. 2 years to reach the market?

Where do they think Baidu, Google, Tesla & Uber will be in 2 years?

> Sure, he worked at Baidu and led a team there, but this is a LOT of money for a very small chance at anything competitive.

There's much more macroeconomics/geopolitics at play on this one and likely more behind the scenes action than anyone here has knowledge of.

Honestly, I don't question what Chinese investors do anymore. They don't play by the same rules as everyone else - they expect lower returns than western firms, they want to get their money out of china and into the US, many of them have political ties, they use capital as a means of status/power (at the expense of financial ROI), and god only knows what is happening behind the scenes in terms of digital espionage.

>> and god only knows what is happening behind the scenes in terms of digital espionage.

I urge anyone who has been working for a large multinational to vanity search for themselves on github/gitlab.

Their main differentiator is that they are China-first.

China doesn't want to miss the boat on any new innovation, and their investment approach is to invest in literally everything, to maximize the likelihood of backing a winner. Tencent bought 5% of Tesla just months ago.

>Their main differentiator is that they are China-first.

And maybe, just maybe, this is connected to:


and to the concept that an autonomous vehicle can be used as a weapon and as such it needs to eb controlled/controllable by a "local" trusted company:


Hypothetically say they have unlimited investment dollars. Can it do harm to invest in everyone?

If you pump unlimited investment dollars into unlimited investments, what tends to happen is that your investees and their employees/suppliers walk off with a lot of money and don't produce any goods or services that result in a sustainable competitive advantage. This is basically the definition of a bubble, and has happened over and over again, from Bitcoin to mobile to subprime mortgages to Web 2.0 to .com to AI to savings & loans. You need some sort of financial constraint - usually the threat of bankruptcy - to create an incentive to invest efficiently, where you make choices about where to allocate resources such that your firm is more competitive than everyone else.

Spreading a small amount of money across a large number of investments within a sector usually works out pretty well, though. This is a diversified portfolio, and basically it means you're betting that someone in the sector will succeed without taking a position on who that will be. If you know that the market exists and the technology is attainable, it makes a lot of sense.

These are private market investments, so nobody is crunching any numbers for them and producing a valuation, they pay for this out of pocket. A small VC would lose with an "invest everywhere" strategy, each new investment grows sunken cost of valuating and negotiating an investment.

But there is a force in the other direction: each investment can only bear so much new capital without losing marginal value. So after a while you have to start to look for new investments to spread your capital around.

There is probably a ton of money behind these chinese investments, and a relatively small total market for venture capital, so they end up making a lot of investments.

> Where do they think Baidu, Google, Tesla & Uber will be in 2 years?

Tesla will be nowhere. They simply do not have the sensor hardware needed to be competitive.

Uber is likely not doing much better[0]. I suspect they will run out of cash before they can deliver a production-ready self-driving car.

The smart money is on Waymo, GM, or Volvo (maybe Toyota).

[0] https://www.recode.net/2017/3/16/14938116/uber-travis-kalani...

I wouldn't discount Tesla:

1. They already have cars on the road, albeit Model 3 has production delays.

2. The people who buy Tesla cars are the same ones who are willing to adopt self-driving technology faster.

Waymo has no brand recognition outside the tech community; GM is old-fashioned, and tailors to alike customers; Volvo and Toyota -- same issue as GM.

1. I wouldn't count Tesla until they make big changes to their hardware and general strategy. They've got an Elon-knows-better-than-the-experts-who-work-for-him problem.

2. Fully autonomous cars will be Robotaxis for the forseeable future. They're not for sale. I'm actually baffled as to how many people there are that don't get this, it's been in the news daily for years. They're far from being idiot-proof enough to put in the hands of private owners, they need careful maintenance and oversight, and cost north of $150k apiece.

GM is moving aggressively, and with confidence. GM acquired Cruise Automation, and they've been developing their Autonomous OS since 2013, it was money well spent. They're the most vertically integrated company making real progress on the self driving problem, and only Waymo is ahead of them. The problem with Waymo is they can't make cars and don't have a legit partnership with anyone who does.

Actually, that's counter intuitive to one Tesla recruiter that I once spoke with. He said that Tesla does not want to take away the driving experience from users which is why having autopilot was good enough for them right now.

This makes sense because without the human driving experience, there's really no differentiation from one self-driving car to another.

You are binning all Tesla owners into just a single category of futurists. In fact there's another group that's in it for the luxury brand and prestige.

Musk is really into the idea of everybody having their own cars. So he came up with The Tesla Network so if you want your Tesla making money for you, you can put your car on it.

He came out with a fabulous idea for a steel vacuum tube that moves shit 700 miles an hour, so you can put your car in it.

He's sick of all the cars clogging up his commute from LA to Fremont, so he decided to buy a boring machine with the goal of a 10x cost reduction in tunnel digging. So we can build awesome subways? No. So you can put your car in it.

SpaceX wants to deliver a preliminary payload to mars orbit and try out the Falcon Heavy. Maybe deploy something useful like a satellite or a rover? To heck with that, let's put a car in orbit!

I don't believe Musk commutes from LA to Fremont. Its actually from his home in Bel Air to the SpaceX HQ in Hawthorne.

I see merit in the Tesla Network because everyone recognizes the potential in an Autonomous Vehicle network. However this will limit the first ideal to have everyone owning their own car. And at the end of the day, once cars become autonomous, it becomes a commodity. People don't really car which brand of the AV gets them to their destination so long as its on time.

>People don't really car which brand of the AV gets them to their destination so long as its on time.

Shouldn't that also imply that people wouldn't care which brand of car gets them to their destination now? That is definitely not the case, for many people.

People care about brand when they are owning, driving, and maintaining them. Probably less so when other people are.

Good point.

Economy class robotaxis will be commoditized, certainly. I don't care what I get to work and back in, just so long as there's a hot coffee waiting for me in the cupholder.

Bob Lutz, former GM CEO and something of a sage in the automotive world takes a sardonic glee in describing a transportation future he probably won't live to see, in which boxy 'modules' ferry us from place to place.

German Car designers, in trying to come up with ways to differentiate themselves in a world where torque and handling no longer matter have placed a focus on the user experience, and have delivered concepts where the interior is reimagined as a living room, with sprawling sofas and wide screen televisions.

These guys are really hurting for ideas. They think of autonomous vehicles as cars that have had the souls sucked out of them. What's there to love? What's there to stir our primitive urges?

But really, it's quite the opposite. They won't be cars anymore. They'll be robots. They'll be animated. They'll be creatures. Alive, and with personalities.

Ford and Dominoes partnered on an autonomous pizza delivery service and were surprised that people often thanked the vehicle for delivering the pizza.

There's a video that has been removed, I wish I could show it to you. Starship technologies is working on sidewalk delivery robots and the prototype looks like a beer cooler on wheels. In the video the bot is trying to cross the road at a crosswalk, and a slowly a car comes around the corner, the driver can't see the bot, it's below his eyeline. The bot, recognizing it's peril, stops and tries to back up, in a last ditch effort for self preservation, but it isn't quick enough, and in slow motion the car runs into it and knocks it over. The bot had a human minder and he quickly righted it and pulled it to the curb. In the comments thread there's all these comments along the lines of 'OMG, that poor little robot, he almost got dided!'. People were anthropomorphizing and empathizing with a fricken beer cooler.

Nobody empathizes with a car when it gets rear ended. We don't project a personality onto it. We're concerned for the passengers, maybe, or the material costs of the damage, but the car has no feelings.

So at the moment we're focused on just getting autonomous to work, it's a technical problem. Autonomous vehicle developers need to demonstrate a minimum viable product that is safe,reliable, and able to generate revenue before they can do much else.

But there is so much else. Once we're ready to start having fun with it, I would trust an artist at Disney to come up with a lucid and original concept for an autonomous vehicle before I would anyone from the tech or automotive industry.

In fact, this is almost the case. Timothy Kentley Klay, the founder of Zoox, worked in digital animation in Australia. Without any automotive or tech industry experience whatsoever he showed up in Silicon Valley with nothing more than a rendering of a bi-directional robotic car he imagined in which the seats face inwards and all four wheels have their own independent steering and motor, and he started knocking on doors and chatting people up in hotel bars at conferences. That was in 2013. Zoox is a unicorn now, miraculously they have $300 million in funding and their test vehicles are out and about in SF and working pretty good.

There's a whole world of possibilities out there that most people, especially car people, aren't thinking of. They're robots of the kind that live and work amongst us, with their own power of agency.

Henry Ford is attributed with saying "If I had asked the people what they wanted, they would've answered 'a faster horse'"

I think it's about time to give the people what they want. Ride 'em, Pony.

No customer wanted to fly a payload on the rocket, so the payload is a joke. The most interesting part about the joke is the small number of people who criticize it as if it were a real problem.

Tesla's SDC program is badly broken. Trust me, or trust the string of engineering departures.

Tesla seems to be focusing on consumer cars. Until sensor prices drop to well below half the cost of a car I can’t see them going all out.

The vehicles that Waymo, GM and Uber are trying won’t be affordable for the average consumer in the short term, and they seem to be focusing on the ride share model.

I think the pitch is that they can be 99.999% sure their vehicle won't kill anyone, vs Tesla being 99.99% sure. If you have enough vehicles on the road, that .009% could translate into quite a few affected drivers. Geofencing constrains the total number of situations to a more manageable level.

Maybe that's how they've pitched this? Otherwise, I can't imagine throwing 100M+ after a late entry to a market saturated with big players with deep pockets.

That, or China gave Sequoia China a great backroom deal.

Not saying it's not stupid, but FWIW I noticed ACRush (competitive programming legend) is behind that startup - at least it should mean it's not a full-blown scam.

I'm reasonably confident in Pony.ai's potential to develop an autonomous OS in China, they've got some real street cred, and in the time they've been at it appears they're coming along fine, possibly they've already surpassed Baidu.(Baidu did a huge restructuring because they weren't getting anywhere).

Pony has an American cousin: Aurora. Aurora was founded at almost the same time by ex-autpilot director Sterling Anderson and former Google SDC project lead Chris Urmson. Just like with Pony they decided they'd be better off starting their own company rather than try to work around the C-suite bullshit they had to put up with at their previous employers.

Now the thing is, top tier talent is short shrift, and the best would rather be working with a Chris Urmson or an ACRush solving hard problems than for a bunch of suits or egomaniacs who just don't get it.

So both Pony and Aurora have a special advantage when it comes to attracting talent that money can't buy. And talent is everything. Many of the best people were passionate about this long before the hype got out of control and the notion of getting stinking rich was ever a serious consideration.

I understand the economic value of self-driving to ridesharing cos but I get this feeling that the car itself is what needs to be "disrupted" and not the fact that a car is or isn't driven by a human. Self-driving doesn't solve the problem of traffic, congestion, pollution (from gas self-driving), etc etc.

I hope hyperloop become reality in the not too distant future and make it feasible/practical for workers living 100mi+ away from urban job centers like SF/NYC/LA to live in distant lower cost communities and comfortably commute into work.

> Self-driving doesn't solve the problem of traffic, congestion, pollution (from gas self-driving), etc etc.

Doesn't it? Looks to me it will (except pollution).

How does it solve traffic/congestion issue? Because they drive in a way that is more efficient than human-driven cars?

I envision many years or decades where having both self-driving and human-driven cars on the road at the same time is going to be a nightmare for everyone...

Also humans regularly drive slightly over the speed limit (at the flow of traffic). I wonder if self-driving cars would be programmed to drive exactly at the speed limit, which means they'd actually be slower.

This is the biggest series A I've heard of. It may be the biggest ever. Admittedly, series A rounds are now de facto series B rounds, but even so, this is very big.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact