By John Gruber
Retool — build native iOS apps with just JS and SQL.
Michael Wayland, reporting last week for CNBC:
General Motors has hired former Apple executive Mike Abbott to lead a newly created software unit for the Detroit automaker. Abbott, former vice president of engineering for Apple’s cloud services division, will join GM as executive vice president of software, effective May 22. He will report to GM CEO Mary Barra.
Abbott’s newly created role will bring together three separate functions within the company: software-defined vehicle and operating systems; information and digital technology; and the company’s digital business. [...]
GM has a target to grow profit margins and double its revenue to about $280 billion by the end of this decade. That includes significant growth in new business units and software.
I’m not familiar with Mike Abbott, but it sounds like his expertise is entirely in cloud services, not user-facing software design and engineering. So I don’t think Barra was thinking “Apple makes great software that people love, so we should hire someone from Apple who knows how to lead such a team”,1 but instead more like “Apple has built an $80 billion per year services division that continues to grow each quarter, so we should hire someone who knows how to lead such a team”. Barra is looking at Apple’s services revenue, not the quality of Apple’s CarPlay.
That’s fine, and maybe someone like Abbott is exactly who GM needs. But I don’t look at this hire and think that GM is any more likely to come up with a CarPlay-quality interface for its own platform. Some back-of-the-envelope math on Barra’s services revenue goals for GM makes it sound to me like Mike Abbott is being tasked with designing an in-dash slot machine.
When GM announced they were dropping CarPlay from their EVs last month, Reuters reported that “Barra is aiming for $20 billion to $25 billion in annual revenue from subscriptions by 2030.”
That seems very ambitious.
Let’s look at GM’s current subscription services revenue. Here’s Larry Printz, writing for Motley Fool last July:
According to company officials, GM generated nearly $2 billion in subscription services revenue and EBIT margins north of 70% in fiscal 2021. The automaker currently has more than 4 million subscribers. For 2021, GM’s global revenue was $127 billion, meaning that if forecast proved true, OnStar accounted for 1.6% of GM’s worldwide revenue. While that may seem like a minimal contribution to the bottom line, that figure should grow thanks to recent additions to OnStar.
GM recently announced a subscription plan for its SuperCruise self-driving feature, which is free for the first three years on new vehicles. It also opened OnStar to owners of non-GM vehicles through a smartphone app, which should bring additional subscribers — and income.
To grow from $2 billion to $20 billion (the low end of Barra’s stated goal) by 2030, it seems safe to assume she’s expecting this growth to come from future car sales, not GM vehicles that are already on the road today.
GM sells about 6 million vehicles per year — but that number has been declining since a peak of 10 million in 2016. Presumably, these future services will require vehicles equipped with GM’s new software platform. Right now they’re claiming this system will only be going into EVs, and in 2022, only 1.7 percent of GM vehicles sold were all-electric.
So let’s be generous and say that by 2030, GM has sold 30 million vehicles eligible for and subscribed to the company’s new services. I think 30 million is very generous — if not outlandish — and would require them to put the new software platform in their gas and hybrid vehicles, too. Keep in mind they only had 4 million subscribers in 2021 and their vehicle sales are in decline.
$20 billion in revenue from 30 million subscribers = $666/year/vehicle = $55/month/vehicle. That’s in line with their current average of $500/year per services subscriber ($2 billion in revenue from 4 million subscribers, per Motley Fool above), but it’s a lot of money per subscriber. I just don’t see how they grow from 4 million subscribers today to 30 million or more by 2030.
What services could they offer that new car buyers would pay north of $50/month for? Maps and navigation? Everyone has that on their phones already. Music and podcasts? Everyone has that on their phones already. Crash detection? By 2030 everyone will have that on their phones already (or at least they will if they have iPhones, but I bet that will soon become a standard feature on Android phones too). GM wants to sell “behavior based insurance” (translation: tracking/surveillance), but according to Reuters, their goal for insurance is just $6 billion/year by 2030. I find it hard to see where the rest of the money will come from.
GM executives might be dreaming that car buyers will pay GM for services they already get on their phones by not supporting CarPlay and Android Auto, but today’s reality shows that people will just buy $5 mounts to attach their phones to their vehicles’ dashboards. And if every other automaker continues to support CarPlay and Android Auto, it just seems like this entire endeavor will turn into an expensive boondoggle that steers would-be GM buyers to other brands.
One idea that occurred to me is the equivalent of Apple’s services revenue dark matter: payments from Google for default placement as Safari’s search engine. Neither Apple nor Google has ever disclosed those numbers, but one recent estimate pegs it at $20 billion per year. At a minimum, it’s $10 billion per year. Those payments from Google go into “Services” in Apple’s quarterly results. Apple now reports $20 billion in services revenue per quarter — so most of that is coming from the App Store, and then customer subscriptions to things like iCloud Plus, Apple Music, etc., but somewhere around 20-25 percent of it comes from Google paying to be the default search engine in Safari. With GM’s upcoming in-house software system not supporting CarPlay or Android Auto, but made in partnership with Google (based on Google’s confusingly-named Android Automotive platform) presumably Google Maps (and/or Waze?) will be the one and only choice for mapping and navigation, and perhaps Google will pay GM for that privilege? And for default placement for other services like music? Even if that’s the deal, though, I don’t see how it gets GM close to $20 billion in services revenue per year. Apple gets $15-20 billion from Google for default search engine placement in Safari on over one billion iPhones, iPads, and Macs. Even my kindest, most optimistic estimate pegs eligible GM vehicles at just 30 million by 2030. Would Google pay GM 30–40 times more per car to provide the default mapping app than they pay Apple per iPhone/iPad/Mac? That doesn’t make sense to me, but I suppose it’s theoretically possible that mapping surveillance data is that valuable to Google. I really fucking doubt it, though.
The one and only thing I can think of that GM can charge a subscription for that drivers can’t get via their phones is semi-autonomous driving. GM calls this Super Cruise and currently charges $25/month for it, or $15/month for drivers who also subscribe to OnStar. (OnStar offers roadside assistance and crash detection — which, again, everyone will have for free with their phones by 2030.) But Super Cruise could exist as a standalone subscription feature alongside support for CarPlay and Android Auto — and in fact does today. Charging a subscription fee for Super Cruise doesn’t require GM to drop support for CarPlay and Android Auto integration.
I can’t see how these ballpark numbers make any sense. It’s an enormous bet against the primacy of the phone in people’s lives, with no one yet, in any industry, ever having won a bet against the phone. GM’s $20–25 billion target for services revenue by 2030 feels like a Kendall Roy number — plucked out of thin air by someone just making up figures in an Excel spreadsheet until they get the
SUM() result they’re looking for.
I don’t think there’s any chance whatsoever this could have happened, but imagine if GM had hired Scott Forstall to lead their software division. That would have been interesting. ↩︎