I’m shutting down InPerson, the 3D telepresence company I started after leaving Apple’s Vision Pro team in 2022. After almost ten years in mixed reality, I’ve come to believe that the industry’s entertainment success is masking a hard truth: we’re nowhere near the next platform shift.
Gorilla Tag hit $100 million in revenue in 2024. VRChat reached 40,000 daily active users last August. The industry looks at these numbers and says “The platform works! General adoption is next!”, but entertainment is a special case, not a leading indicator, and understanding why is the reason I walked away.
After the mobile revolution of the 2010s, Big Tech has been searching for the next platform shift. Platform shifts, when new technology and new distribution methods fundamentally change what’s possible to build, flip the competitive landscape and not only birth new players, but destroy incumbents. This is exactly the reason why Meta, Apple, Google have poured billions underwriting the future by funding mixed reality efforts.
Platform shifts follow a pattern:
We’ve seen this before. Personal computers became practical for home and office use in the late 1970s as they became affordable and small enough for a desk. Only then did spreadsheet software emerge (VisiCalc, 1979), creating the “killer app” that made PCs essential for business. The iPhone followed the same arc decades later.
But if we already see fervent, daily usage of mixed reality for entertainment1, why haven’t we seen those killer apps yet?
Entertainment success doesn’t actually prove we’re ready for a platform shift. It only shows that mixed reality works for low-stakes, session-based use cases.
Entertainment users tolerate discomfort because fun is immediate and sessions are short. Developers can design experiences within the 20-to-40-minute “Goldilocks zone” to minimize fatigue. General computing has the opposite properties: work isn’t fun enough to justify discomfort, tasks can’t be compressed into short sessions, and every use case competes directly with laptops and phones that are far more comfortable. In other words, entertainment success tells us nothing about all-day, high-stakes general computing viability.
And yes, niche professional use cases like surgical training and architecture walkthroughs work, but precisely because they’re high-value enough to justify short-duration discomfort.
The iPhone followed the platform shift pattern. Practical hardware enabled daily use, which unlocked general computing, which let killer apps emerge. But the discovery took time. Uber launched in 2009, two years after the iPhone, only viable because GPS, payment systems and always-on internet enabled a workflow impossible before. Instagram launched in 2010, emerging because mobile cameras, social graphs and persistent connectivity created entirely new social behaviors.
These apps emerged once the iPhone made sustained daily use possible, even if widespread adoption was still growing. Developers and users needed millions of hours of real-world usage to uncover workflows that took advantage of the platform’s unique traits.
Mixed reality hasn’t gotten there. The hardware isn’t good enough for all-day use; headsets are simply too uncomfortable. We have yet to achieve step one of the platform shift pattern: comfortable hardware that enables daily use. The industry is rightly trending towards lighter headsets, but we’re at least two generations away from all-day wearable, capable devices2.
When we get there, the killer apps will likely surprise us just as Instagram surprised everyone when the iPhone made cameras ubiquitous.
I left Apple’s Vision Pro team back in 2022, hoping to catch market tailwinds in 3D. I had been keeping a pulse check on real-time 3D facial reconstruction research and figured it was only a couple years away. It felt to me that there was a window opening, where I could productionize research that was nearly there (my specialty) with devices that were able to render 3D content. It felt like the perfect moment to solve the “holy grail” of mixed reality: meeting someone virtually as if in real-life, as if in-person.
What I didn’t expect was for us to still be at step one. Though over the last four years InPerson has been able to produce stunning, impressive demos of hyperrealistic avatars, the problem remains that no one actually owns a headset. We couldn’t acquire users because there were no users. What I thought were general computing devices were still gaming devices; not quite there for the masses to use. We had excited beta testers and advanced partnership talks with 3D display companies and AR/VR startups who wanted the tech. But our potential partners had the same problem we did: no users.
I’m not the only one who made this mistake. Meta lost $70 billion since 2020 (!) trying to bring us to step four but it turns out the hardware problem is just very difficult and Moore’s Law isn’t fast enough. Remember, this is not to say that mixed reality won’t pan out. It’s just that it hasn’t in the past five years, and probably won’t in the next five years.
I could have raised more money and kept going. But founders are supposed to ride through hype cycles and winters. The problem is, in four years I never once felt what market pull even feels like. That’s not a winter. That’s a signal.
The biggest lesson for me is no matter how well or how badly you operate your startup (admittedly as a first-time founder I didn’t operate it well enough), it actually doesn’t matter if the market doesn’t exist.
Thanks to Nat Friedman, Daniel Gross, and the team at NFDG who bet on this vision alongside me, and to Audrey for reading drafts of this. Also, I wrote most of this article on an Apple Vision Pro.