Loading today’s stories
Product3 distinct publishers3 min readUpdated
Meta shipped Pocket the day Gizmo died, and never had to buy anything
Meta hired Gizmo's team in March, licensed the technology without buying the company, and launched the result nationally the day the original went dark. Nvidia paid $6bn for the same shape of deal.
The Product Desk · Product desk
What happened
- Meta launched Pocket across the United States, and the app it is based on shut down the same day.
- Pocket had been testing quietly in Brazil since July and is now available to everyone in the US.
- Pocket lets people type a prompt and get back a small interactive thing, which Meta calls a gizmo; gizmos publish to a scrollable feed where other people can play, save, repost or remix them.
- Gizmo was built by Atma Sciences, a company founded by former Snapchat engineers.
- Meta did not buy Atma Sciences. It hired the team and took a non-exclusive licence to the technology, Business Insider reported, having scooped the move in March.
Compiled by The Product DeskSomething wrong?How this is made
Why it matters
The load-bearing word in this deal is non-exclusive. Meta took a licence it did not need to hold alone [6], which only makes sense if the code was never the scarce asset. The people were, and they left in March [6]. Whatever sat at Atma Sciences afterwards was technology a third party could in principle license and nobody was left to operate, so Thursday's wind-down notice [8] reads as bookkeeping rather than an event. Josh Siegel, who ran it, called the shutdown bittersweet and said he was excited for Pocket [9]. Both of those can be true at once.
Nvidia's version of the same structure supplies a price. It agreed to pay Poolside $6bn for a licence and jobs for 109 staff, with both sides insisting this was not an acquisition [10]. That works out at roughly $55m per hired head [12]. The Next Web reckons Meta's version cost less, and it plainly ran faster: team in during March, product shipped in July, the original gone by August [11], five months from hire to shutdown and four from hire to launch [13]. TechCrunch calls it an acqui-hire [21]. Business Insider, which has used both apps, says the two experiences are pretty much identical [7]. Nobody filed to buy anything.
Then there is what Pocket runs on. Muse Spark comes out of Meta Superintelligence Labs [14], and outside developers spent months waiting on an API that Meta kept pushing back, the model having launched in April with the interface trailing it [15]. Meta has not said where that now stands [16]. What is on the record is the order of operations: the first consumer product shipped on the model is Meta's own [16].
The distribution choice is the part that does not fit the house pattern. Meta told Business Insider it has no plans for now to connect Pocket to Instagram or the rest of the family [17], which is not how this company normally launches anything. Pocket's spread instead depends on gizmo links opening for people who have not installed it [18], with comments, likes and reposts inside an app that has no direct messages [18]. Zuckerberg told analysts in July that AI-assisted development is what makes shipping new apps easier, and that Meta intends to build more ideas and scale the ones that work through its recommendation systems [19]. Pocket joins Instagram Instants, Forum, Seller, a bedtime-stories experiment, a Creator Studio app and a Mac client for Meta AI [20]. That is a portfolio built to be pruned, and an app with no connection to the family is the cheapest kind to prune.
Engadget's read was that Pocket is worth a download and a couple of minutes before boredom sets in [22]. If that is where it lands, the durable artefact is not the app. It is the deal shape: four months to ship, no purchase price, and the app it was copied from switched off on launch day [1].
Loading topic controls
Loading related stories