At this year’s Meta Connect, Zuckerberg previewed a world where an AI companion works in the background to “make you smarter, help you communicate better, improve your memory, improve your senses, and more”.

What that means is Meta is effectively reframing the metaverse concept away from VR headsets toward AR wearables. Instead of expecting people to enter virtual worlds, Zuckerberg’s metaverse vision now centers on “ambient computing”, real world experiences enhanced by AI and lifelike 3D visuals. In practice, Meta wants to provide fashionable AR glasses that people will actually want to wear.

This means glasses could overlay translations, directions, or holographic call participants into your field of view, making digital interactions feel face-to-face. Notably, this approach treats the metaverse as a blend of “helpful AI plus shared experiences,” not purely immersive VR worlds.

He considers glasses are considered the best form factor for AI because they’re heads-up and hands free: you can stay engaged with your surroundings while getting contextual help that boosts your memory, communication and senses, rather than staring down at a phone.

My Take: Zuckerberg’s pivot from a VR centric metaverse to AI powered wearable computing is a notable strategy shift. He’s effectively saying the true promise of the metaverse will be realized when personalized AI and everyday glasses, rather than solely in virtual reality headsets. 

The challenge is delivering enough utility to convince someone with perfect vision to wear glasses all day, which is a steep behavior change. History offers mixed evidence: the smartphone was adopted quickly (went from 40m units a year in 2004 to 122m in 2007) because it delivered huge value, but other wearables have grown slower, including smart watches. Globally there are about 100m smart watches sold a year, ten years after the launch of Apple Watch. This compares to about to 1.3B smart phones sold annually. Most people don’t enjoy wearing glasses for long periods unless the utility warrants the effort. 

Putting it together, I think we’re five years away from hitting the 100m annual glasses shipments.

One final thought: Zuckerberg talks about these glasses allowing us to be more present. In fact, any notification you get in your glasses are basically the same distraction as if they came over the phone. You’ll still glance at something, just in a different place. That means they might reduce time looking down at a phone, but you’re still looking at a screen, just a transparent one.

While glasses are the hardware side of this future, AI is the brains behind it. Zuckerberg’s vision involves an AI that is personalized for each user, effectively a social companion, creative assistant, and info guru tailored to you. He believes that Meta is distinct from other AI players because it focuses on “AI that’s your personal superintelligence” rather than a one size fits all model.

The idea is that your AI knows you and can pull in context about your interests, friends, and routines to give relevant answers, becoming a personalized social assistant. Zuckerberg suggests that just as smartphones became indispensable, these AI assistants will become core consumer tech.

I expect most of the advancements in personalized AI to be powered by agents. One example Zuckerberg showed was an upgraded agent capable of generating full virtual worlds inside of Horizon Worlds. This has little appeal to most today, and will have a lot of appeal if the metaverse catches on. I brainstormed for more tangible examples of how Meta’s personalized AI agents can make our lives better including:

• Schedule meetings
• Book appointments
• Coordinate group events
• Compare prices across retailers
• Edit social photos and videos
• Recommend products
• Track delivery status
• Book fights based on budget, dates, and preferences.
• Reserve hotels
• Plan travel
• Monitor and alert for flight or hotel price drops
• Review utility, cable, or mobile bills for errors
• Negotiate or provide scripts to lower bill costs
• Generate weekly meal plans based on diet and budget
• Review and cancel unused subscriptions

In simpler terms, Meta is trying to build AI helpers that feel less like generic chatbots and more like your own digital sidekick that “gets you”. It not only has general knowledge, but also knows who and what matters to you. Meta’s advantage here is the trove of social data. This personalized AI would live throughout Meta’s products and accessible on a phone, computer or wearable (likely won’t need a Meta wearable).

My Take: While true personalized AI from an assistant that genuinely understands your context, preferences and relationships is a compelling goal, we are still three plus years away. Right now, AI assistants (Siri, ChatGPT, Grok) don’t know you personally. They respond with general intelligence and perhaps some basic user info. Meta’s plan hints at an AI that continuously learns about you and can act on your behalf in the digital world. If done right, it would offload tedious tasks, help us make better decisions, and enhance social.

The bottom line is that a truly personalized, human level social AI is still a work in progress, and we will eventually get there, like it or not. This will fundamentally redefine how we interact with tech and by extension, how we interact with each other.

Earlier this summer, Zuckerberg didn’t just hop on the superintelligence train, he became its conductor. He got to work outlining unprecedented Capex spending plans and equally unprecedented billion dollar pay packages for top AI talent.

Taking a step back, superintelligence is the concept that machines eventually will be able to think at an abstract level beyond humans’ ability to understand. The thinking is that superintelligence will follow general intelligence, which we are still a few years away from reaching. The timing between general and superintelligence could be as short as months or as long as never. Safe to say, the concept of superintelligence falls in the radical yet doable AI roadmap and Meta is all in.

At Connect Zuckerberg said Meta is “working to bring personal superintelligence to everyone”. By using the term superintelligence he is setting a very high bar, given it suggests an AI that is beyond just smart; one that gives each person capabilities exceeding genius.

One way at Connect that superintelligence came up was how Zuckerberg talked about the glasses and AI together: “Glasses are the ideal form factor for personal superintelligence,” he said, because they allow an AI to see and hear everything you do and provide help in real time without pulling you out of the moment to grab your phone.

Between the lines, Zuckerberg is telling us that Meta intends to be at the forefront of the next AI revolution, not just building chatbots but creating tech that is woven into our lives that can help us beyond what we can understand today.

The level of investment the company is making around AI, and therefore towards superintelligence, is two standard deviations ahead of the rest of big tech. For example, next year Meta expects Capex, which is an index to how much they invest in AI, to be up 47%. That compares to Google, Amazon and Microsoft, which have suggested their spending will be on average 7% next year. While I expect the rest of big tech to step up Capex outlooks on the September earnings calls, they will likely still lag Meta’s level of investment.

Additionally, comments from Zuckerberg on September 4 at the White House, followed by comments from CFO Susan Li at the Goldman conference on September 9, underscore that aggressive Capex investment will continue into 2027 and 2028. Adjusting for operational spending in the US over the next three years, which should account for about $120B in total, the company implied it will spend about $480B in Capex over the next few years. That suggests Capex will grow at 47% in 2026 and 45% in 2027 and 2028. It is safe to say these numbers are hard to believe, and I expect despite the comments from Zuckerberg and Li, the actual Capex growth will fall short of that 45% growth threshold. In the end, I expect the out years to increase at closer to 20% per year, which I still see as a resounding statement of the company’s determination to see superintelligence become a reality.

My Take: Meta’s pursuit of personal superintelligence is as ambitious as it sounds. On one hand, I admire the audacity: Zuckerberg is willing to bet the farm on a big idea, much as he did when pivoting to mobile in 2012 or to the metaverse in 2021. Meta has shown that it is ready to back this vision with enormous resources as outlined above. This level of spending and commitment gives Meta a chance to will advanced AI into existence that other companies might not gamble on. If any company outside the very largest, Google and Microsoft, can chase true superintelligence like capabilities, Meta is in that group by virtue of its talent and budget.

Disclaimer