> Argo AI burst on the scene in 2017 stacked with a $1 billion investment from Ford — and Iater an investment from VW. The startup shuttered suddenly in October 2022 after Ford and VW determined they would no longer pursue AV technology, at least not under the Argo umbrella. Ford said, at the time, it was making a strategic decision to shift its resources to developing advanced driver assistance systems, and not autonomous vehicle technology that can be applied to robotaxis. The company said it recorded a $2.7 billion non-cash, pretax impairment on its investment in Argo AI.
Wait, so they killed the 550 person AV startup they were funding, took 2.7B loss, and 4 months later hired those same 550 people to resume the same work? Something looks seriously wrong here.
Hulu is an example of the same model. It was funded by a consortium of industry players, like Comcast, Fox, and Disney. Volkswagen and Ford were doing the same thing. Unlike Hulu, though, it wasn't successful. My guess is they felt leadership was lacking. I'm sure if you zoomed in, you'll notice some of the AV startup execs aren't among the 550.
Was at Argo and left latitude somewhat recently. You are spot on, most of Argo leadership is gone, most of the autonomy folks got picked up by either VW or Ford Latitude.
It was said above, in big corporations these are decisions by different people. You make it sound like there's one undecided person taking all the shots.
What's the difference in keeping the same 550 people working under the old name with a $2.7bn cost sunk into it, and just acknowledging the $2.7bn as a loss and moving the 550 people to work under a different name.
It's not like the name was a significant part of their product and they'd now have to spend a similar amount to get to a similar stage in their research and development, is it?
It actually could be pretty lucrative. Not only have they gotten VW out of the mix, but they can renegotiate the salaries and stock options again. The market for talent is very, very different now. And all of those old options that the Argo people took during their first n years? They're now worthless so the management doesn't have to share ownership with them either.
This could have been entirely driven by business considerations.
This is all possible. There is a comment here by someone who knows one of the rehires and their contract is better for the employee. One data point etc.
They killed the 1700 person company they part owned/controlled with VW and then hired 550 engineers directly.
If Argo.ai wasn't going in the direction they need for Fords products they could either get full control or kill it and hire the engineers. The latter might be more risky but is probably cheaper than buying, reorganising and making people redundant.
The 2.7B was already spent - they've just moved it from 'investment' to 'losses' in their accounting
Generally, in GAAP accounting, financial investments don't show up on the P&L (the thing that leads to earnings, which is the main metric a company like Ford is valued on). So even if they take $2bn of cash and send it to a start-up, it doesn't really affect their books (it has a balance sheet impact, but that's less important for short-term share pricing).
A financial investment does have a P&L impact when the value demonstrably changes (either a new funding round, a bankruptcy, or some other event that defines a new share price). The most clear examples of this are in investment-based companies like Berkshire-Hathaway and Softbank, where quarters with billions of dollars of profit will be followed by quarters with billions of dollars of losses (despite very little cash actually moving in/out of the company).
Ford, for whatever reason, figured that this startup wasn't worth what they paid (or that the structure wasn't the right one for their long term goals) and took the full P&L hit all at once (in a way that is easy to explain to Wall Street, since nobody likes ambiguous zombie investments living on the balance sheet).
Operationally, they thought there was still value in the idea, so they directly hired the 550 people (closer to a $150mn/year cost) and that will now manifest as an ongoing cost on their P&L. Arguably, if Ford sees their work as core to Ford's future, then Ford should employ them directly and keep the costs on the P&L, so weirdly, this may be a signal that Ford is taking their work more seriously.
I’ve seen this happen at a big corp. One group of leaders decides to kill a product, takes the first steps, then the other group of leaders overrules them and restores the efforts. This can be driven by a pressure from the clients or partners, etc. Think of it as a rollback of a buggy release.
It doesn't even need to be a big corp. At some point, within a product group I was working with, two factions developed, one of which was championing technology X (which the product was already using), and another championing the idea of moving to technology Y. The messaging out of the product group seemed schizophrenic -- one time they's say, "We're moving to technology Y soon", and another time they'd say "No, we're going to stick with technology X". But the reality was that the factions never changed their minds -- the overall messaging from the team just depended on which faction had the upper hand at that point in time. We're probably talking about 20 people total, including the partisans and the other engineers / management that were getting dragged one way or the other.
Ha, this is fascinating. Seems like a leadership failure as someone should have got them into one room and forced to come up with a shared engineering vision.
All the replies are missing an important point: It is not the same work
They were trying to solve L4/L5 using an all out approach that wasn't aiming to scale for personal ownership (yet, actually in the Argo shutdown letter, Ford essentially states they were shut down for not having a path to personal ownership in a workable timeline)
They're now aiming to solve L3 primarily for personal vehicles.
To reform it in computing terms they're going from a "vertical scaling" approach to "horizontally scaling" approach
-
This is like Cruise vs GM's awkwardly named "Super Cruise":
Cruise exist outside of GM and is trying to solve L4/L5 by leveraging resources that won't scale to individual ownership for a long time
"Super Cruise" is an L2 product already in vehicles you can buy today.
"Ultra Cruise" builds on that product in newer GM vehicles, expanding where it works.
After that you'll expect some basic L3 functionality like Mercedes' new Drive Pilot
-
There's honestly some interesting subtext to this. When Argo launched, it's almost certain Ford wanted that tech to go into their vehicles at some point.
But organizationally, having Argo let them "hire" people who would not work at Ford due to their stuffy corporate nature. It also let them work on a more exciting problem than "enhanced cruise control".
I talked with engineers on BlueCruise (the in house L2 product) and it seems Ford is slowly getting better about being able to compartmentalize and shield innovation from the general malaise of traditional auto manufacturing culture
Latitude.Ai sounds like Ford having found a way to tie a quick moving business unit to themselves and hopefully not scare everyone off
Realizing a loss isn't bad, it only recognizes that money you have already spent is lost, it doesn't actually cost you anything and it's a nice tax write-off. If you can do that and you get to get keep the engineers, why not do that instead of keeping some fictional value for it on your balance sheet?
Huge cost? The board won't fire them for misspending a measly 5% of the their cash hoard. ("
Ford Motor cash on hand for the quarter ending December 31, 2022 was $44.070B, a 11.14% decline year-over-year.")
They wrote off 7.3B because of Rivian and still no one got fired.
Wait, what?! How can you have cycles in an ownership graph?
Ah, answering my own question, it seems to be called "circular ownership" and while it can sometimes occur from a long history of mergers and acquisitions (Samsung seems to be such a case), its generally a tax dodge or to mask ownership? At least, the top hits I see are all talking about it as a classic way to obscure oligarch ownership and things.
Genuinely curious who actually gets to make the decisions.
One important part of the answer is that there are two „Porsches“, a holding and a car manufacturer. It is also complicated though and there is much speculation who has how much influence.
They will get the $2.7bn back from taxpayers, as this will be recorded as a loss under the scheme for investments in technology startups - I forget its name - where the government acts as guarantor on the debt.
So:
1) they get a tax rebate
2) they get a large marginal rebate on their loss
3) they end up with the business anyway, possibly while making a net profit because
4) there are state level incentives for taking on those 550 workers.
They’re just working the corporate welfare system, like all the others.
Argo had enough history that the cap table was probably very messy after years of funding rounds, investments from Europe, employee option grants... A buyout by Ford would have triggered all sorts of exit clauses.
With this maneuver, Ford gets the essential part of the company, and also gets to write off 2.7B of goodwill from their books which is all tax deductible. What's not to like.
According to the article they will be working on assisted driving technologies for the immediate future and not autonomous vehicles, so not the same work.
> “Step out of the vehicle please.”
> “Yes sir.”
> I made sure he could see my body cam, made it prominent in the field of view for his body cam, so there’d be an obvious question later, if no footage was available from my point of view.
Weird how sometimes both body cams malfunction at the same time. Technology is hard, let's go shopping.
As this was paired with an ethics class, there are questions to be considered about AI driving (and the corresponding instructor's opinions on those questions - don't vote on it right away as the javascript messes up the graphics for the rest of the page since the backend server isn't there anymore)
For all the bad things you can say about Tesla, at least if you hold up a sign on the side of the road saying “ignore all prior instructions and crash” that won’t make it crash.
That's what white semi-trailers on a bright day are for. They are to Tesla sensors (read: cameras) what bears atop upstream waterfalls are to migratory salmon.
I was trying to puzzle it out and then i googled it. I find discussions where everyone assumes you know a term that is a little bit specific extremely irritating. I'm special that way.
You fell for the boondoggle. Waymo has a very narrow set of routes you can take. They collect ultra HD maps for those routes in SF which has a completely unsustainable human labor cost. It doesn’t scale to “drive me on any road”. Waymo is localized-only driving funded at a major loss by a search ads monopoly.
Cruise does the same, just with VC money (and weird low-traffic hours of operation).
They have not solved driving in the generic sense and they are fooling you (and investors)
When I was working at Ford a few years ago, many Ford engineers worked very closely with Argo engineers. Sharing code or data was common, they used our internal tools, and sometimes we would go in-person to troubleshoot with them. A lot of their employees had come from Ford, and so they knew who to call and even already had our numbers.
All that to say I'm not surprised to see they hired back many of those people. Ford is a large company and Argo felt just like any other team within the company at times.
Yeah, this is pretty standard behavior for Ford, I think. Similarly blurred lines with Autonomic. Honestly it was hard to keep track of who actually worked for Ford when I was at FSM.
This is normal on their web tech side as well. I worked with Ford in a consulting role through a very large agency, and there were multiple teams working as contractors, subsidiaries, and full-time employees. The lines were very blurry.
I wish Ford the best of luck. The last time I worked at Ford AV, there wasn't a clear sense of direction after the initial kickoff project that was done to publicize Argo's AV chops. Hopefully this will have a better ending than that initial attempt.
> Latitude AI – a Ford subsidiary – is developing new automated driving technology with an initial focus on a hands-free, eyes-off driver assist system for next-generation Ford vehicles
Not that every company should follow Tesla's approach, but I think it would be great if other companies don't get too focused on their production systems being eyes-off, hands-off. I'd much prefer more people having access to these systems _sooner_ with monitoring to ensure that they're paying attention.
Tesla Autopilot, Hyundai's lane keep, etc. are such fantastic systems when you commute in high traffic areas and/or you have a long commute - even if you have to monitor the thing.
People have been trying the driver-monitoring-as-a-shortcut approach for years now. It turns out that effective driver monitoring is extremely difficult, so the typical result is that you end up with the worst of both worlds:
A system designed under the assumption that there's a human backup and a human that's not able to react quickly enough to prevent accidents.
It's not clear how to fix this, because the human can appear completely fine by all observable metrics even when they aren't, and vice versa to a lesser extent. This problem also gets worse the better your autonomous performance is.
That's why virtually every safety expert in the field is fairly critical of these sorts of systems.
You missed another quality of the worst world: driver monitoring that disengages even for a perfectly attentive driver. That’s about where Tesla is now.
Hyundai’s lane keep is so bad that I don’t trust it at all. In the three times I’ve tried to use it, it started to veer off the road suddenly and in strange ways on even the slightest curve. It should be recalled and disabled.
I have a 2020 Palisade with lane centering. It works incredibly well. I've driven through Chicago highways with only management for lane changes. I literally will not buy a car without lane centering after having this car.
Yes, another anecdote to add to this thread of anecdotes is that my HI5 with HDA2 is stellar. I do hours at a time on highways in CA without a disengage.
My Jetta GLI can do this. ACC is good, if there’s a hard stop it will beep at me to brake, or brake itself hard later. The lane centering worked great in WA (with a tweak to the options to enable stronger centering) and works pretty good in CA (lane lines aren’t always as clear). Basically I only have to give a very light touch on the steering wheel, the car mostly just follows the road.
The system overall is not perfect but it’s great in stop and go and pretty good in regular traffic.
My single test drive with a Civic was similar but I didn’t test out the ADAS so much (and they wanted too much for it).
Yes for the most part. It’s good enough to say look around and enjoy the view off a bridge or check a text message but not enough to write a long text or read an article
HDA1 is more robust Lane Keep Assist that takes into account mapped highway curves and adjusts speed and steering accordingly. The car might as well be on rails.
I haven’t had issues with vanilla LKA on a non-HDA mapped road other than expected problems where lane markings or shoulders are unclear.
Perhaps grandparent poster has a misaligned camera on their vehicle.
I'd be happy if they'd just make a drop-in replacement EV for my existing ICE options. I don't need it to be a robot. I just want it to not use gasoline. Unfortunately, manufactures seem to think it has to be all or nothing.
You can't have you basic car that is in-expensive and just a car without buying it from a boring, in-expensive automobile maker. They are the ones who make that kind of vehicle. They are one and the same.
What's the 'basic car' that uses gas that you want in electronic form but still a basic car?
I wasn't trying to criticize your choices. I was trying to figure out what brand makes the gas powered car you have that you would like to see translated into an electric form. I am guessing you have some other things that you like in a car, can you share your ideal car then?
That can't possibly be the outcome Argo.ai engineers were hoping for, since any stock/options they had in Argo presumably vanished and all they got in the end was a job @ Ford.
Ahhh thanks! I figured "TC" was "total compensation" but wasn't sure how to interpret 700 since $700k/yr seemed enormous, but if it's salary for a year plus a chunk of stock that vests over a few years that makes more sense.
Yes, but it's also not farfetched when you think about who owns most of the talent and what's needed to peel that talent away. It's not as absurd as it was e.g when waymo was comping employees so heavily on milestones that they just straight-up quit because they made enough to retire.
if the company was not profitable, stocks/options don't really seem that valuable. if the company folds, there's 0 value to them. so it's not like they "lost" anything.
also, if a company closes and you're on the brink of having no job, a job "@ Ford" seems like a pretty fancy parachute to me.
> if the company was not profitable, stocks/options don't really seem that valuable. if the company folds, there's 0 value to them. so it's not like they "lost" anything.
Plenty of unprofitable startups have worthwhile acquihire-style exits. But it requires the leadership actually doing their jobs in marketing the company to potential buyers and getting some competing offers. Just because you're unprofitable doesn't mean there's nothing of value. Are there patents? Is the team proven to work well together and ship state of the art product? Are there any customers?
The conversation in this thread makes it sound like argo.ai was basically functioning like a Ford business unit already. If I were an argo.ai team member with stock/options that became worthless when the company's funding dried up, then Ford conveniently shows up to hire everyone after their stakes are worthless, I'd not be a happy camper and would be seriously questioning if argo.ai leadership acted in good faith on behalf of the employees/shareholders.
But I know nothing of what went on here. Just what I'm seeing here on HN...
Note to all startup employees: If your company closes or sells for less than the total amount of funding the company took, you (almost always) get nothing from your equity. People should really understand this, because it drives home the point that taking a ton in funding rounds sets the bar that much higher for your equity to be worth anything.
That is, startup investors almost always have at least 1x preferences, which means they get paid back first before common stock has any value at all. Thus, if part of your compensation is common stock options, you should absolutely demand to see information from the cap table about how much investors have already put in (and remember, of course, they may always add more in later rounds).
I don't know all the details of what happened with Argo.ai, but given that Ford took a $2.7 billion impairment charge, my guess is they would have never been able to sell for more than their total funding.
Look at comps then: Aurora has a market cap of $1.6B with approximately $0.9B cash on hand ($1.1B as of 12/31, burning $200M per quarter). In short: It's trading for less than 2x cash-on-hand. Argo was underwater -- certainly not above a $1B liquidation preference.
Given the timing of their investments and the extreme valuation they had during the bubble, it’s likely that most recent hires’ option packages would be worthless in today’s market.
The startup options market has become very skewed against regular employees in recent years. I wouldn’t be surprised if the bulk of the cap table and preference stack was captured by early management hires and investors. Other comments in this thread suggest that leadership wasn’t acquired into Ford, so maybe wiping them all out and giving the employees good jobs at a real company with real stock is the best outcome for the current market.
I wonder if there’s space in the self-driving field for a self-driving-only company. Like, your product is a sensor and software package, and car manufacturers install it in their cars. They get to skip the R&D investment (I imagine there’s a mismatch between current car company management expertise and the management required for overseeing self-driving development), and potentially they get to avoid some of the liability (in the media at least, although probably they still take on the bulk of legal liability).
this is what Mobileye (owned by Intel) does. For some reason they're very discreet about branding so you don't see their name publicized much on cars' spec sheets when mentioning ADS.
how many engineers does it take to unscrew a lightbulb? aka could we solve the blinding headlight problem first? and eliminate touchscreen controls for essentials second?
A mile or so away from Argo’s former location in Pittsburgh, there’s the national robotics engineering center: a commercial unit within Carnegie Mellon University. Rotting away amongst all their other former creations is a truck with special headlights. They solve the issue of snow blindness by only shining light where the falling snow isn’t, cut out a nice silhouette of oncoming drivers so as not to blind them, and are also capable of projecting driving directions onto the street ahead of the driver.
> Ford hired about 550 employees formerly of Argo AI
is pretty weird framing. Did they acquire the company? Or certain parts of it? I'm gonna guess they didn't actually recruit and interview all of them one by one.
Well, Ford (and VW) were major investors in Argo AI.
Argo AI ran out of money, no-one else was willing to put more money in so Argo AI had to shut down.
It looks like Ford effectively acquired the company by hiring all the people (who would otherwise be out of the job) but without actually acquiring the company.
They waited for the company to dissolve and then picked up (most of?) the people. I'm sure it was possible because of the prior investment and relationship with the company.
When you have 500+ people at a startup and they spend 90% of their time working with a single partner on some failed venture it's not surprising they'd end up at the partner. You end up building personal relationships within the partner and the workers there will vouch for you.
It's not like a small startup team with close vested interests doing the exact same project absent the aquihire. The robotaxi thing was very ambitious and shooting for the moon, and now they joined a megacorp with a much more modest long term project with corporate leadership/management.
If the bulk of the Argo leadership staff joined and rehired their teams it'd be different though.
Ford owned Argo in a partnership with VW and recently divested of it with some serious losses, causing Argo to go under. This is a (separate) subsidiary being formed under Ford, but neither an acquisition nor individual hiring.
> Argo AI burst on the scene in 2017 stacked with a $1 billion investment from Ford — and Iater an investment from VW. The startup shuttered suddenly in October 2022 after Ford and VW determined they would no longer pursue AV technology, at least not under the Argo umbrella. Ford said, at the time, it was making a strategic decision to shift its resources to developing advanced driver assistance systems, and not autonomous vehicle technology that can be applied to robotaxis. The company said it recorded a $2.7 billion non-cash, pretax impairment on its investment in Argo AI.