How Meta’s AI & VR Strategies Are Aligning At Last
A short update on Zuckerberg’s Metaverse long game
Meta’s Busy Fortnight
It’s been quite a busy fortnight for Meta. After teasing that it is set to unleash its Llama 3 open-source large language model (LLM) within a month at the beginning of April, the company delivered the goods last week, releasing a compact version of Llama 3 and reaffirming its commitment to the open-souce model. In a blog post, Meta claimed Llama 3 beat similarly sized models like Google’s Gemma and Gemini, and Anthropic’s Claude 3 in certain benchmarking tests.
In tandem with the release of Llama 3, Meta also rolled out an updated Meta AI, its virtual assistant now present across Meta apps like Facebook and Instagram, which can now answer questions using real-time information from Google and Bing. From testing Meta AI in WhatsApp in India to experimenting with AI in Instagram’s search bar for both communication and content discovery, Meta is putting its AI chatbot front and center in the user experience on those platforms.
This week, Meta turned its attention to its mixed reality business. First, the company announced that it is rebranding Quest OS, the operating system for Oculus Quest VR headsets, as Meta Horizon OS, and making it open to third-party manufacturers including the likes of ASUS and Lenovo. Apparently, it is also making a “limited-edition Meta Quest, inspired by Xbox,” although it is unclear whether Microsoft will be directly involved in its design or manufacturing. This move marks a strategic pivot for Meta’s VR business, as it concedes total control over VR hardware and aims to make Horizon OS into the Android of VR instead.
Second, the company is also adding Meta AI to its smart glasses made in collaboration with Ray-Ban, thus allowing them to gain multimodal AI functions such as identifying a plant or pet, or processing information from an image or audio recording. Meta CTO Andrew Bosworth shared his thoughts on AI on Meta’s hardware in a recent interview, in which he suggested that generative AI features could appear on Meta’s VR headsets soon.
Why Meta Chooses Openness
It is interesting to contrast Meta’s open-source stance regarding its LLMs from the get-go with its newfound openness in regard to its VR software. Partly acknowledging the competitive pressure in the current tech landscape, and partly to fulfill Mark Zuckerberg’s long-held platform dreams, Meta is trying to position itself as a neutral partner willing to work with everybody as it seeks to reap the benefits of community-driven innovation. Moreover, by adopting a more collaborative open-source approach, Meta also gets to preemptively fend off regulators that are increasingly turning the antitrust screws on Big Tech.
The open-source nature of the Llama models, while technically limited to applications with under 700 million monthly active users, opens up opportunities for a broader range of developers, especially smaller companies and independent developers who don’t have the resources to develop their own LLMs or to license top-tier models. It is, in fact, in Meta’s best interest to have as many good-enough LLMs out in the market to counter the closed LLMs from OpenAI or Google. Meta doesn’t necessarily need to have the best AI models; it just needs to make sure it doesn’t become dependent on ChatGPT or Gemini down the road.
Similarly, opening up the software ecosystem for its VR headsets also benefits Meta in two ways. First, Horizon OS will allow more VR manufacturers to create headsets that run on, and are presumably integrated with, Meta’s Horizon Worlds — the work-in-progress foundation for the company’s lofty metaverse ambitions. Last September, Meta expanded Horizon Worlds to allow more people to access its social VR experience via their phone, and via desktop PCs. This move may help bring more people into its fold via third-party devices.
Second, VR has been a rather stagnant market for a while, and Meta, as basically the sole leader in the VR headset category at this point, needs to drum up more excitement for the category. Opening up Horizon OS would encourage more gadget manufacturers to take a stab at building VR headsets, and it is in Meta’s best interest to encourage competition rather than trying to gatekeep.
Taken together, these moves by Meta represent a concerted push toward building out key components of the metaverse — AI, mixed reality OS, and, most importantly, adding AI to its services and hardware — as open and interoperable layers. Aligning its VR strategy to its open approach to AI, which Zuckerberg claimed had replaced the metaverse as Meta’s top priority in 2024, paints a fuller picture of the type of long game that Zuckerberg is playing.
The endless creative power that generative AI has unleashed will be crucial to the world-building of its metaverse platform, and Meta recognizes that. Already, Instagram is reportedly testing Creator AI, a program that leverages Meta AI to allow influencers to chat with fans via customized direct messages — essentially allowing influencers to clone themselves as a chatbot that is available 24/7. Add that on top of the celebrity-based AI chatbots that Meta has already rolled out, it is clear that AI characters will play a part in the future of Meta’s social platforms.
By open-sourcing AI models, opening up its VR OS, and working across hardware partners, Meta is painting a vision of the metaverse as a community-driven, AI-assisted creative project. Whether this open strategy can truly counterbalance the walled garden approaches of rivals like Apple remains to be seen. While the Vision Pro is clearly going after a different market segment than Meta’s Oculus headsets at the moment, they are ultimately competing in the same product category of immersive experience headsets.
As analyst Ben Thompson sums it up in his insightful column on Meta and its seemingly open source approach:
This remaining limitation does get at the rather fuzzy meaning of “open”: in the case of Horizon OS, Meta means a licensing model for its OS and more freedom for developers relative to Apple; in the case of Llama, Meta means open weights and making models into a commodity…
Meta, in other words, is not taking some sort of philosophical stand: rather, they are clear-eyed about what their market is (time and attention), and their core differentiation (horizontal services that capture more time and attention than anyone); everything that matters in pursuit of that market and maintenance of that differentiation is worth investing in, and if “openness” means that investment goes further or performs better or handicaps a competitor, then Meta will be open.
What Lies Ahead
Meta’s timing is pretty good this time around, considering the looming “TikTok ban,” which President Biden just signed off on, and the recent bellyflop of Humane’s AI Pin upon launch. Theoretically, there are (or likely will be, in the case of TikTok) vacuums in both sectors to be filled, and Meta happens to be the only company with their hands in both markets.
Yet, despite everything making sense strategically on paper, Meta would still need to convince its investors first. On Thursday, Meta’s stock shares plunged by 16% following a weak revenue forecast, despite the better-than-expected Q1 results. On the earnings call, Zuckerberg boasted that “tens of millions of people” have “tried” Meta AI, but he also frankly stated that it will be a costly endeavor and that any hope of making money from Meta AI is likely several years away.
All these concerns point to a short-term push to further monetize user attention on its social platforms. Already, AI also increasingly underpins Meta’s advertising infrastructure through offerings like the Advantage+ suite of products and an ads-ranking system called Meta Lattice.
Still, Meta is clearly committed to building its open source strategies as key pillars for its future growth, especially as it aims to transform into a more diversified tech company not solely reliant on social media revenues. These recent developments around Meta AI and Horizon OS signify a broader move by the company to align its technology stack with its vision for the metaverse, where AI-assisted creations and interactions will power the ever-expanding immersive virtual experiences.
Only time will tell whether Meta can pull it off.
