Welcome to Lowpass! This week: The latest on Amazon’s secret new TV OS, and Meta’s plans to jumpstart its metaverse development with AI.

Amazon continues to pursue a transition away from Android for its TV hardware: The company plans to release a first TV streaming device powered by its still-unannounced Vega OS later this year and has been courting major publishers to bring their apps to the platform, I’ve been able to confirm with sources familiar with the company’s plans as well as through a number of leaks.

An Amazon spokesperson declined to comment.

Vega is a Linux-based operating system that may one day replace Amazon’s Android-based Fire OS across a wide variety of devices, as I was first able to report in November of 2023. The company has already launched three Vega-powered devices: The company’s Echo Show 5 and Echo Hub smart displays both run Vega, as does the Echo Spot smart clock / speaker.

Amazon has been slower in adopting Vega for its smart TVs and streaming players. The company originally planned to launch a Vega-powered streaming stick in late 2024, but ultimately delayed that launch. More recently, Amazon revealed that it is bringing an updated version of its Android-based TV OS to smart TVs released this year, leading to some speculation that the company may be abandoning Vega.

That couldn’t be further from the truth, as demonstrated by a growing number of leaks.

Amazon’s own job site currently lists multiple positions openly mentioning Vega, including one published this week for a “Software Development Engineer, Vega OS” who will be joining the “Vega Runtime” team. “You will design, implement, and maintain low-level services that ensure seamless app execution, enforce strict security policies, and optimize system resources across a range of devices, including Multimodal, TV, and Automotive,” the job listing explains.

Amazon is also currently looking to fill roles for a SDET II, Vega TV, a System Dev Engineer, Vega OS and a Quality Assurance Technician I, Vega support. (The company changed some of the job descriptions on its own site to remove references to Vega after I reached out.)

This week, the company also published a job listing for a Software Development Engineer, Kepler SDK. Kepler is the name for Vega’s software development kit, which allows developers to build React Native apps for the platform – the name of the SDK hasn’t been previously reported. 

Amazon has been making an aggressive push to establish React Native as the default application framework for TV app developers, with developer advocates regularly speaking at React Native industry events. The company also recently teamed up with React Native-focused developer consultancy Callstack to release an ebook about React Native app development for TV, and hosted a meetup in New York to promote both.

The company has yet to include references to Vega in this public outreach to developers, but I’ve been told that the pitch gets a lot more detailed behind closed doors: Callstack and Amazon also jointly held an invite-only TV app developer summit at Amazon’s headquarters in Seattle in January, where they courted major app publishers for Vega. A since-deleted description of the event promised attendees updates on “the latest Fire TV features, SDKs, and launch initiatives.”

I’ve been able to confirm that Paramount, Rakuten and the BBC’s UKTV subsidiary are among the publishers building TV apps using the Kepler SDK. Some other developers appear to be more hesitant to adopt the new platform, and are opting to port their HTML-based TV apps instead. Amazon does not currently offer a way to directly port Android-based TV apps to the new platform, I’ve been told.

This could lead to a situation similar to Amazon’s transition from its existing Alexa voice assistant to its new LLM-powered Alexa+ assistant: The apps of most major publishers are expected to be available on the new platform, but a lot of long-tail apps may not make the cut.

Meta is making it easier for developers to create Horizon Worlds environments: The company announced the launch of new generative AI tools this week that are supposed to cut down development times from weeks to just hours.

The new tools, which include the ability to create meshes and textures with generative AI, are part of the recently-launched Horizon desktop editor – a Windows app to create 3D worlds for the company’s metaverse platform. Earlier this year, the company added the ability to create Skyboxes – the 3D spheres that surround VR apps – with generative AI.

With this week’s updates, that ability is extended to the actual objects and environments players can explore and interact with. A demo video published by Meta shows a developer creating a grass-covered hill and some rock formations by entering a simple text prompt. Developers can also use AI to generate Typescript, the scripting language used by Horizon Worlds to add behavior to objects, as well as sound effects and ambient audio.

Meta’s embrace of AI for Horizon Worlds comes as the company is looking to accelerate growth of the platform, both in headset and on phones and desktop computers – a subject that Meta Reality Labs Director of Games Chris Pruett touched on during his recent GDC talk.  

“When it comes to Horizon Worlds, our goal is to make sure that it grows much quicker than it’s currently growing,” Pruett acknowledged during that talk. “It’s actually had a great year, but it needs to grow a lot faster.”

Key to growing the platform’s audience quicker is bringing Horizon Worlds content to mobile, Puett argued. Meta began making some Horizon games available on mobile phones in 2023, but Pruett acknowledged that those efforts are still far from complete. “I would consider the current stuff we have as pretty early and mostly a test, but we’re shipping and learning from it pretty quickly,” he said, adding: “In the next year, we’re going to dramatically improve the content and delivery quality of Horizon Worlds on phones.”

“This is a huge major focus for us,” Pruett said.

Last week, it finally happened: I accidentally sent a version of the newsletter featuring a sponsorship message to paying subscribers, who are supposed to get this newsletter ad-free. Turns out I didn’t click the thing. Apologies!

Speaking of which: Lowpass has had some great advertisers recently, and more in store for the coming weeks and months. Their sponsorships allow me to chase scoops and report on the things that matter when it comes to the future of tech and entertainment, and I hope you find value in them as well. If you’d rather read Lowpass ad-free, I encourage you to sign up for a premium membership.

And if you’re a company looking to reach decision makers in streaming, AR and VR: More than 95% of this newsletter’s readers are on the free, ad-supported tier. Get in touch to book your Q3 sponsorship slots now.

Patreon gets into live video. The membership services company is making the ability to go live available to select creators this week, and plans a formal launch this summer.

Apple is working on two new VR headsets. The company’s plans reportedly include a cheaper and lighter Vision Pro successor, as well as a headset that’s tethered to a Macbook.

Netflix is testing AI-powered search. The company is testing new search features powered by OpenAI in Australia and New Zealand.

Black Magic nixes plans to make cameras in the US. Not exactly how these tariffs were supposed to work: Faced with potentially sky-high import taxes on components, Hollywood camera maker Blackmagic Design decided building a factory in the US is unsustainable.

18% of the music uploaded to Deezer is now AI-generated. Wowzers: 20,000 AI-generated tracks are being uploaded to Deezer every day.

Netflix wants to be worth as much as Apple. The streaming giant wants to double its revenue by 2030, and reach a market cap of $1 trillion, according to the Wall Street Journal.

Google is adding video generation to Gemini Advanced. Paying users of Google’s Gemini AI tool can now use it to generate short clips with Google’s Veo 2 AI video generator.

Advertising could be hit hard by trade war. A majority of ad execs expects revenue declines of 6-10 percent — which could be bad news for ad-supported video, among other things.

Sooner or later, everything will be augmented. Case in point: Austin-based connected fitness startup Growl has built a projection-based AR boxing coach. Imagine a punching bag that has a virtual coach projected on top of it, complete with motion tracking. Sounds neat, but it’s also not exactly cheap (the company eventually wants to sell these things for $5000 each). Maybe someone should develop a mixed reality boxing game that incorporates real-life punching bags for those of us with tighter wallets? Or you could just go to the gym, I guess …

Thanks for reading, have a great weekend!

Photo by sajjad soltanpoor on Unsplash