Hacker News Reader

Comment by ams6110 | Hacker News Reader

Honda had an existing partnership with GM to develop fuel cells. The Detroit automaker has its Chevrolet Bolt EV in showrooms and has developed a battery pack that will be used in 20 new EVs globally by 2023. In June, the companies announced a deal that would have Honda using GM’s cells and battery pack. GM’s battery not only had good range, but by working together the two automakers figured they could lower costs.

The thing about companies like Honda and GM -- they have a lot of momentum. That means they aren't likely to be on the forefront of innovation, but when they turn towards a new target, they will hit with massive resources and established global production capacity that a start-up can only dream of.

View on HN

>The thing about companies like Honda and GM -- they have a lot of momentum. That means they aren't likely to be on the forefront of innovation, but when they turn towards a new target, they will hit with massive resources and established global production capacity that a start-up can only dream of.

Exactly. It's the difference between a handful of dudes in a speedboat with some explosives and an aircraft carrier. Once the latter is aware of the former you can be reasonably assured who of will win. At most the former can hope to land a good sucker punch before the latter knows they're even fighting.

The core competency of GM, Honda, etc. is building cars. Once they decide what kind of cars they want to build they will do a good job building them as they intend to build them. They don't make rookie mistakes, especially in the logistics of spinning up production lines to build what they want to build. They've re-tool their factories every time a new generation or a new model comes out. They know how to do it right the first time, knowledge that comes from experience.

They are good at building cars, but I would argue they are not good at building tech. Self driving is a technology issue. If the tech superstars at Waymo/Apple/whoever aren't even close to solving it, why would the software people at GM/Honda be?

"Hi, this Jane from the procurement department of GM North America. We might be interested in licensing your software. Please give me a call back so we can set up a meeting."

They don't need to build tech. They can buy it. GM can write some pretty big checks. They flat out bought Cruise just to cover their butts.

When I was an intern at a defense contractor ~1/10th the size of GM my cube was across from procurement. I think a lot of people here don't understand the kind of scale companies like this play at. Your problem isn't "how do we pay for this" it's "will anyone sell me all this all at once or sign a contract to provide it all over the timetable we want". You make bets on how many times around the earth the welding wire you'll buy next quarter will go. GM shipped ~10 million vehicles in 2017. That means they had to sign a contract with someone to supply approximately an Exxon Valdez worth of motor oil and another of ATF just for their factories.

Isn't that exactly what a deal is?

This wouldn't be a loss for Waymo if everyone was licensing their software.

Waymo is far the best self driving tech and nobody has the money to buy it, as Alphabet is not selling (it's the first real diversification from ads revenue). GM is few years behind, but hopefully those few years are not enough for Waymo to create a monopoly.

GM really isn't far behind.

Self Driving is not only a technical problem, it will be a people/marketing/legal/political problem. By the time these will be solved, GM/Tesla/Toyota should have competing solutions.

Exactly. Non-Waymo companies have the benefit of Google/Waymo taking care of many of these obstacles for them.

But they have the problem of getting solutions adapted to Waymo's strategy. They'll likely push for safety tests and requirements that they can pass; others will have to adapt to that.

> as Alphabet is not selling

They will license/sell it eventually when/if they will get it to work. What else would they do with that tech ? Last time I've checked Alphabet did not produce any cars or had any infrastructure to do that on scale.

I think cloud is google's first real diversification from Ads. Nobody looks at Waymo as a revenue center.

You have too short a time horizon on your "nobody".

Waymo is Google’s first AI robotics spinoff. That’s supposedly a (young) growth market.

Supposedly. I am skeptical that Waymo will ever execute a major deal. It's a real shame, the tech is really good, but Waymo just isn't very good at business (this is true about Alphabet in general; beyond ads, and maybe cloud).

XXX is not very good at business, except <several billion dollar product> and <several billion dollar product>.

They have tried and failed multiple dozen times to create new products. They're bad at business.

You could say maybe they’re bad at startups but good at growing established products.

There is truth to it. A friend of mine used the HUGE opportunity Alphabet os missing by failing to offer a service that allows to buy from vendors through Google only ising your Goovle account. Combine that with dropship and you have a potential Amazon-Killer in your pocket.

you're talking about Google Pay and Google Express

Maybe, my friend referred to the search results under shopping. But that's Europe.

Looking at the current news cycle, I think their huge lead is eventually going to get squandered away. George Hotz said Waymo is like Xerox PARC of Autonomous Tech. I think it's somewhere in between.They're somewhat like Fairchild Semiconductor. A lot of their great engineers have gone on to found their own companies. Be it Anton or Chris. Tesla already has the brand firepower on its side, to unknowingly create the perception that they're self-driving tech pioneers.

"but Waymo just isn't very good at business"

They can hire the right people but the problem might be trust. IMO, no major car company wants to deal with Google, they'll be enslaved to Google who will tighten the screws little by little.

Perhaps overlooked is that Waymo also is coming at self-driving from a very different angle than the car companies. Waymo, based on everything they've said and done, is focused on full autonomy even if that takes quite a few years to roll out broadly.

The auto makers are more than fine with incremental tech so long as it sells more cars. In some ways they may actually prefer it to the degree it accelerates upgrade cycles. Yes, there are issues with intermediate ~ Level 3 stages. But, in general, the car companies don't really care all that much about full self-driving and, arguably, would actually prefer if it hits the streets later rather than sooner.

I'm in total agreement. I told everybody at Waymo (back when it was part of X) that the car companies are fully incremental- they roll features out over years and push them down from the luxury models to the cheapies. They would far prefer to gain knowledge through incremental improvements each year.

The Waymo folks absorbed this, agreed, and said they would proceed towards full autonomy- I kind of get this, for example I don't even use cruise control and don't plan on using L3 autonomous because I worry about losing attentiveness and getting in an accident.

Whether (and you say 'arguably') they are actively preferring to slow down the rollout of full self-driving is hard to say; I think they're just cautious and don't want a few dramatic events to scuttle this promising area of research.

If they can crack that nut, they’ll have a crazy moat for a coupl years. And if they scale like hell in that moment, they might have a real brand on their hands.

And there’s no guarantee at all that the incremental approach will lead to the same place in a reasonable timeframe.

So, risks all around. And people playing to their strengths.

Having the best tech isn't a predictor of success. Just ask BetaMax and LaserDisk...

What matters more is the business strategy, and articles like this are suggesting that Waymo's business plan will keep it from dominating the market.

It turns out that trying to compete with your suppliers is very, very expensive, especially if you try to make them shoulder the risks without sharing the upsides. And very expensive business plans require very good execution, which Waymo hasn't managed so far. They've already been rejected by 3 automakers (Ford, GM, and Honda).

They don't need to be interested in selling off Waymo, because as we see in this very article they are interested in licensing Waymo tech to established players like Honda.

Funny that Google did this very thing to become the search and advertising giant it is today.

Wait so who is actually going to build the technology?

Cruise Automation presumably, which is owned by GM.

Cruise Automation was cofounded by Kyle Vogt, a YC alum who cofounded, built Justin.TV, which became Twitch, and sold to Amazon for a billion dollars before starting Cruise with Daniel Kan, which was acquired in early 2016 by GM. It's a tech company. It employs it's share of tech superstars. Vogt himself participated in the Darpa challenges as a robotics undergrad at MIT, before dropping out to do Justin.TV.

The only logical outcome is that the big car companies continue to build the majority of cars when (if) self-driving cars are ready for the mass market. Waymo is not going to build millions of cars a year like GM and Ford and Toyota.

Big car companies can buy startups in the self driving space, which is what GM did with Cruise. Or they can license the software from Waymo or other companies who are working on it.

Self-driving software can be just another thing car companies buy from suppliers, like transmissions and brakes.

I think you are oversimplifying.

'Tech' as a commonly applied label typically involves widgets and software. Not large manufacturing and supply chains that involves world trade, weather, regionalism, human resources in the hundreds of thousands and years of planning to work with many similarly configured partners.

Tesla for all it's glory isn't the greatest endorsement for Tech as some (not necessarily you) like to claim and I like Tesla.

Everyone that says car companies don't move fast seems to be picking and choosing the facts that suit their own bias towards 'traditional' industries.

Car companies only look slow on the outside, but they really aren't when you look at the totality of their products scope.

For example car companies go through millions of dollars in just testing alone with nearly every country operating their own testing center. They need to be weather tested in so many different environments, they need to meet regulatory hurdles that are different from state to state to State. It's just massive what goes into what they need to do just to get the right to sell their product in a region or country.

Yet they do all this with significant upgrades to models that are now coming nearly every 3 years. I remember when it went from 10 years to 5 years. Wow!

I remember when the assembly got so complicated that cars were squeaking due to the materials rubbing against each other. So in comes material science guys trying to get that sorted in all weather conditions.

I think Waymo is far more likely to be in the sidelines than people think, but we'll see; I could be wrong.

I fully agree. There's tons of stuff moving fast when you look on the inside. Everyone aware of what the cutting edge is in their field and working to implement that within their own little niche of responsibility. From the outside it's like watching a large military cargo plane. It's so big that it looks slow no matter how fast it's going.

That said, GM (or any other BigCo) realizes when it doesn't have a capability and either decides to not pursue it, spin it up themselves internally (like in the case of EVs), license it, or buy someone who has it (e.g. Cruise).

>I remember when the assembly got so complicated that cars were squeaking due to the materials rubbing against each other. So in comes material science guys trying to get that sorted in all weather conditions.

Was that back when you could look at a picture of a Saturn and tell what temperature it was when the picture was taken? ;)

They are tech, just not silicon valley tech. More engineers per capita in Michigan etc.,etc.

I would vastly prefer to drive a car that's been tested by engineers in Michigan rather than the precious folk in Silicon Valley.

The software people at GM aren't a part of GM's existing tech groups. They went and bought a (YC-backed) self-driving startup, Cruise Automation, to get this effort going.

I think the combination of Cruise's expertise on self-driving tech combined with GM's on cars makes for a pretty serious competitor, assuming the two can work together effectively.

>It's the difference between a handful of dudes in a speedboat with some explosives and an aircraft carrier.

That’s uh... one helluva metaphor you got there.

It's a somewhat infamous tactic used by a US leader in a war game simulation of war with a hypothetical mid east enemy (Iran of course), he used first a wave of cruise missiles, then a follow up attack of many small boats on suicide runs to overwhelm a carrier force (hypothetically) and won a two week planned exercise in one day, so the folks in charge re-ran the simulation and barred the tactics he used. https://en.wikipedia.org/wiki/Millennium_Challenge_2002

True that, if could with the pants down a carrier is nothing more than a big target. There is alo an example (no source, sorry) of a German sub surfacing next to a US carrier, within the inner defense ring, suring an exercise.

But all that only works of the carrier is caught flat footed. To stay in the metaphor, the carrier "automotive industry" is no longer flat footed. Partol aircraft and choppers are deployed and the defenses are up.

By the way, I like that image, fits the business world much bettwr than the proverbial tanker vessel for big co.

Sweden produces short range diesel electric subs that were able to sink a US carrier in a similar war game so the US Navy leased one to learn how to adapt to the previously unknown weakness. IIRC China was making these kinds of subs but they have pretty short range, compared to most attack subs.

The point of a simulation is to find vulnerabilities. When something works the correct thing is to rerun it with the assumption that you have mitigated that attack to see what your next vulnerability is. You of course also make sure the mitigation actually happens.

Lawyers run practice trials all the time, if their opponent in practice comes up with an argument they can't destroy on the spot they know where they need to focus.

Yes, they should have at least tried once again with no restrictions to see if the assymetric warfare worked with the Blue team prepared for it by the first run experience, the barring the tactic totally seems to be the opposite of what they should have done.

I had a reply written for that poster, but decided against posting, but yeah, look up the USS Cole. The ship was ruined, people killed. Not talking about some military exercise, but about reality.

Edit: wow, still reading comments, everyone talking about exercises and not about when a speedboat actually went up against a warship. Maybe a perfect analogy, all theory, no real world with some crowds.

> I had a reply written for that poster, but decided against posting, but yeah, look up the USS Cole. The ship was ruined, people killed. Not talking about some military exercise, but about reality.

I think the GGP had that covered:

>>> At most the former can hope to land a good sucker punch before the latter knows they're even fighting.

The USS Cole attack was the sucker punch. Now that the US Navy knows they're fighting, they reprogrammed their robot guns to attack speedboats:

https://www.navy.mil/navydata/fact_display.asp?cid=2100&tid=...

https://www.youtube.com/watch?v=izP3liVmfCE

It's necessary to have the sentence following even try to make sense.

>> Once the latter is aware of the former you can be reasonably assured who of will win.

What does it even mean to "win" in the context of the EV market? Or even just the car market generally? Markets may be competitive but that doesn't translate into the competitiveness of a battle. It's a wonder the world finds a place for variety of so many products and goods at all, given people's obsession over "the winner", somehow creating a monopoly, and fear of losing some "battle" that isn't actually happening.

The thing is - self-driving cars fundamentally change the structure of the automobile market in a way that negates many of the advantages of existing automakers. Instead of private ownership of cars, they encourage ride-sharing services, because the car can pick up a different passenger on the return trip and no longer has to park. Instead of internal combustion engines, they encourage EVs, both because of the power-draw of the self-driving hardware and because refueling becomes the responsibility of the service, which can program them to return to base and not pick up any new passengers once they're getting low on charge. Instead of servicing being a huge pain that encourages buyers to put a premium on reliable cars, maintenance is handled by a centralized maintenance department full of professionals. Instead of family cars, they encourage small motorcycle-like "pods" sized for a single occupant (with a smaller number of family cars or trucks available for family trips).

Already, that negates the advantages of consumer brand, reliability, dealer networks, internal combustion engines, ergonomics, and bodywork. There's not all that much left to the auto industry, once all that changes.

Even if you are right with everything you predicted (some of it, all of it is doubtfull), that does not negate the core competencies of traditional car makers.

You still need someone to buold cars -> existing car makers

EVs are actually easier to built than EVs -> inceeasing the advantage they have on production

Maintenance -> to be done locally, there is no way to centralize it on a full continent, car makers and their existing network are king

Fleet management -> see above, plus existing fleet mgt serives, only real competition are large rental car companies

Ride sharing -> acquisitions, mytaxi in the case of Mercendes, drivenow and car2go in the case of Mercedes and BMW, VW with an in-house solution

Self-driving -> acquisition of the provider (GM and Cruise) or sourcing the tech, alternative is a partnership (Jaguar)

Traditional car makers slept into these developments, a few years ago I would have bet a lot on disruptors in that industry. Now that changed, the only "tech" (as in software) companies that can compete are the likes of Google and maybe Uber as long as VC money doesn't run out. The same is true for certain smaller manufacturers of automobiles, but this market cleaning is gling on for while now so they are used to it. And let's fave it, the automotive sector is well entrenched in politics (Europe, Japan and the US, Chine, too) and money generally is not a rwal problem. So, yeah, now I would say Google / Wymo might become a major Tier one supplier. But hardly more.

When you large fleets of cars you will do maintenance more like how cars are built. With basically an assembly line.

But you would also push some of the work out to cheaper labor and centrally.

So replace the entire wheel instead of the tire. Then ship off the wheel with the old tire to be replaced somewhere else and centrally.

But it is about the economics. Today car dealers lose money on every car sold and make their money from maintenance.

“Yes, the typical new car sold loses a dealership about $200.”

https://www.cargurus.com/Cars/articles/show_me_the_money_how...

Which is why maintenance is so expensive. I would expect Waymo to invest heavily on improving this significantly.

Waymo will not be a tier 1 provider, IMO. They have shared rolling out their commercial ride sharing in Arizona later this year and testing today.

The thing is they started way earlier then others. This is from nine years ago.

https://www.youtube.com/watch?v=4V2bcbJZuPQ

Google was the top place to work for six straight years and so they were getting the cream of the crop. They then get to triage those engineers and send the cream of the crop of the cream of the crop to Waymo.

Then there is the massive infrastructure that Google has to leverage.

SDC is an software engineering exercise. Who is better at software engineering?

Look at the automated testing that Waymo has.

https://www.theatlantic.com/technology/archive/2017/08/insid...

How would you ever do SDC without this?

Great post and completely agree.

Today cars are sold through dealers. Those dealers lose money on every car sold.

“Yes, the typical new car sold loses a dealership about $200.”

https://www.cargurus.com/Cars/articles/show_me_the_money_how...

They make their money from maintenance and why maintenance is so expensive and so common.

Waymo or any other company running very large fleets of cars would fix this.

They would take the service data each year and invest to improve. Plus you have perfect data on how the car was used with SDC.

Car designs will also change. Today you have to design around the person being able to see when driving.

No longer needed.

How is the power draw even a factor for the fuel? A sedan is using 10 or 20 kW just driving around, 2 kW to power a computer isn't going to require fundamentally rethinking anything, just a correctly sized alternator (or whatnot these days).

That electricity is cheaper is the big factor, especially for a service doing shorter runs. Whether it's used for traction or steering doesn't really matter.

FWIW, Honda builds more two-wheeled vehicles (scooters and motorcycles) than it does cars. About 19 million motorcycles a year vs about 5 million cars.

Their core competency is probably engines, especially small engines. They make more engines than anyone else globally.

(I agree with your point generally on momentum. Just saying because I was somewhat surprised to learn it.)

Reply on news.ycombinator.com