Welcome to Lowpass! This week: A second look at (and life for) Spotify hardware, and Meta’s CTO tells me why he is rallying the troops for 2025.

🎅 🎁 🤶 🎁 Still need a last-minute present for the AR, VR or streaming insider in your life? Lowpass subscriptions are now giftable!

What hacking Car Thing taught me about Spotify hardware

The other day, it finally happened: Spotify killed my favorite way of using its service on the go – its Car Thing display.

This didn’t come as much of a surprise. After releasing Car Thing as its first hardware product in early 2022, Spotify quickly changed its tune. In July of 2022, Spotify announced that it would discontinue Car Thing; earlier this year, it warned that existing Car Thing devices would cease operating in December.

Spotify told users like me that they should recycle the device when that time had come. Instead, when mine stopped working this past weekend, I hacked it, and installed an alternative app framework on it that allows me to run it as a Spotify controller on my desk.

Doing so, talking to one of the developers behind this hacked version, and reading through a bunch of eulogies former Car Thing developers have posted on LinkedIn in recent days has given me a new appreciation of Spotify’s complicated hardware journey.

/sub

How Car Thing became a thing. Car Thing was one of two hardware products incubated by Spotify’s “Daytona” team, with the other being the aptly-named Home Thing smart speaker that ultimately never got released. Car Thing also went through multiple iterations; an earlier version had onboard cellular connectivity, and basically functioned as a standalone device. 

Spotify ultimately settled on a thin-client-like approach, with Car Thing connecting via Bluetooth to both your phone and your in-car stereo system, which allowed the company to bring down costs. Still, Car Thing packed some impressive technology: In addition to a touch screen, the device featured a jog dial, a home button and five programmable shortcut buttons. Also on board: Multiple far-field microphones optimized to recognize song requests in spite the kinds loud background noises you have to deal with in a moving vehicle.

Why Spotify built  Car Thing. When Car Thing launched in early 2022, it was billed as a way to bring music streaming to older cars without connected dashboards. That was definitely the primary use case for the device, but the development of Spotify’s hardware projects did not happen in a vacuum.

In the automotive Space, Car Thing was meant to preempt Apple’s and Google’s CarPlay and Android Automotive initiatives. More broadly, hardware was a way for Spotify to bring its own voice assistant to the world, which itself was a hedge against Siri, Alexa and the Google Assistant. Music services execs have long complained about the fact that these third-party voice assistants give them limited visibility into what users actually want. If Alexa misunderstands a user’s voice request of a song for another voice command entirely, Spotify is never going to know. And if Siri defaults podcast requests to its own service, Spotify is left empty-handed.

Spotify closely tracked the performance of its own voice assistant on Car Thing, with the company’s former hardware product lead Anders Arnqvist noting this month that Car Thing “outperformed the benchmarks of Amazon Alexa, Google Home, and Apple Siri by far in both wake word performance and latency.”

Why Car Thing got killed so quickly. When Spotify announced its retreat from hardware two years ago, it cited both “product demand” and “supply chain issues” as reasons. Former employees echoed those sentiments on LinkedIn in recent days, with a senior PM writing that the product “didn’t find [a] perfect market fit,” but that “the pandemic, component crisis and supply chain disturbances [also] played a part.”

Perhaps not entirely coincidental, Spotify announced its retreat from hardware around the same time it pulled the plug on its own voice assistant in its mobile apps.

What Spotify is losing without hardware. By giving up on its own hardware plans, Spotify is not just ceding ground to big platform operators like Apple, Google and Amazon. Giving up on the input modalities of a device like Car Thing also risks making the service more bland, according to Arnqvist.

“Car Thing catered really well to lean-in listening through its multimodal UX — multiple options for steering, pivoting, and following spur-of-the-moment whims,” Arnqvist wrote.  “In data, we saw an increase in user activity, meaning more signals to feed algorithms that generate recommendations. Now, with the increased use of AI products and AI-based recommendations, I worry that the increased lean-back behavior will gradually erode the value of those recommendations and their ability to inspire and help with discovery.”

How hackers are trying to rescue Car Thing. Spotify’s May announcement to brick Car Things by the end of the year motivated a bunch of tinkerers to find alternative uses for the display. Among them: A student who goes by the name Riprod online, and initially had something else altogether in mind for his Car Thing.

“It started just as a funny way to show a Trello to-do list,” Riprod told me recently. Once he got that working, he kept tinkering, and with the help of others built what eventually became Desk Thing – a way to turn Car Thing into a desktop media controller and widget player. A Spotify-forward smart display, if you will.

Desk Thing is just one of a number of Car Thing hacks, and it’s definitely a hack at the moment: To run it, you need to install a desktop server and keep the device plugged in via USB at all times. What’s more, Desk Thing is still very experimental. In my case, it was able to display song names, but failed to load album art – a known bug when connected to a Mac. 

Riprod wants to eventually power Desk Thing by a React Native app, which would allow it to once again function in the car, connected to a mobile phone. So far, he and other members of the Car Thing hacking community have gotten little to no help from Spotify itself, which has left them without access to a few key device components. “Microphone access, proximity sensor access or Bluetooth access would be a big thing,” he told me.

Perhaps, Car Thing hackers will one day get some help from the very people who built the hardware in the first place. Reminiscing about Car Thing, one of them wrote on LinkedIn this week: “I still have my Car Thing, and I could never part with it. Maybe it’s time to explore a new use case for this hardware – what do you think, Daytona team?”

Meta’s CTO: We’re not alone in AR/VR anymore

Meta CTO Andrew Bosworth penned a year-end post about the company’s AR/VR efforts this week that doubled as a kind of rallying cry for Meta employees, with Bosworth declaring that “the coming year will be the most important one in the history of Reality Labs.”

I caught up with Bosworth after the post went live, and asked him if he thought 2025 would be a kind of make-it-or-break-it year for Meta in AR and VR. “I don’t know about breaking it,” he retorted, while acknowledging that the stakes are high. “We are leading in these spaces, and the competition is coming for us.”

“I did bring that message to my team: There’s real stakes now,” Bosworth added, “If we slip, then we potentially lose our leadership position.”

Meta’s newly-arrived competition includes Apple with its Vision Pro headset, as well as Google with its recent Android XR announcement. “Up until now, we were a little bit on our own,” Bosworth said. “We were having to advance the state of the art across every single part of our portfolio.” Now, others are swooping in, and benefitting from some of the work that Meta has done in the past. “Competition always has an easier time than the leaders,”  he said. “The leader is pathfinding, and it’s expensive. The competition is just following what the leader does.”

Apple executives and Apple Vision Pro fans would probably disagree with that characterization. However, but during our conversation, it became clear that Bosworth is paying just as much attention to companies working on possible competitors to the company’s Ray-Ban Meta smart glasses. “You've already seen a ton of knockoffs coming out of China,” Bosworth said. “We know that many robust enterprises are also looking at the space.”

“I joke with my team: We did all this amazing work to establish ourselves in a leadership position, and the reward for that is competition,” Bosworth quipped.

Bosworth stressed that the past couple of years had prepared Meta well for this new competition, and pointed to the company’s experience in rolling out AI on its smart glasses, as well as building metaverse experiences for Quest headsets, as proof points.

Regarding the latter, Bosworth revealed this week that Horizon Worlds is now among the three most-used immersive apps on the new Quest 3s headset. “It's up there with YouTube and Gorilla Tag,” he told me. Bosworth teased what he called “an exciting roadmap” for Horizon Worlds for 2025, and said that some of his team's work would focus on making more Horizon Worlds experiences available on mobile phones. 

At the same time, Meta is also continuing its work to make Quest headsets useful in work-related contexts – an area that Apple has put a bigger emphasis on, and that Samsung seems to be zeroing in on as well with its Project Moohan headset. “I think we're going to be incredibly competitive in the same exact use cases that they're targeting,” Bosworth said.

Meta may get some help on that front from third-party device makers like Lenovo and Asus, who have signed up to develop their own hardware running Meta’s Horizon OS. Bosworth didn’t share any updates on those devices with me, but said that working with device makers required some internal changes as well. “When you build an operating system for first-party devices, you can take a lot of shortcuts,” he acknowledged.

One thing is already clear: 2025 will be a busy year for AR/VR. Samsung is poised to release its headset in partnership with Google, Apple expected to release a cheaper Vision Pro as early as next fall, and the smart glasses space is becoming more competitive. 

Bosworth told me that he wants his team to be ready. “I want to make sure we exit next year feeling like we're still in the pole position [in] these categories,” he said. “We created these categories. We were working on them when everyone else was laughing at us. Telling us that they thought it was crazy, and decreasing their investments. And now, even though they are coming for us, I want to stay in the front.”

What else

Fubo’s lawsuit against Venu survives in court. Sports streaming service Venu will have to further postpone its launch, as its backers failed to get Fubo’s antitrust lawsuit dismissed.

Amazon ditches plastic for its device packaging. The company is also using less paper and ink for the packaging of its Fire TV, Kindle and Echo devices.

Ozy Media co-founder Carlos Watson gets hefty prison sentence. Watson was sentenced to almost ten years for defrauding investors.

Google announce new video generation model. Veo 2 looks super impressive, especially since Google’s model appears to be better at physics than OpenAI’s Sora.

Big Bird is looking for a new home. Max has not renewed its Sesame Street contract, which means that the show could pop up on a different streamer soon.

That’s it

Congratulations! You’ve made it through the final Lowpass newsletter of the year. I’ll be taking off the next two weeks, and will be back with a new issue during CES week. Until then, enjoy your break, and happy holidays!

Thanks for reading, have a great weekend everyone! charging