Y CombinatorApple Vision Pro: Startup Platform Of The Future?
CHAPTERS
- 0:00 – 1:04
Vision Pro as a new founder platform (not a review)
The hosts frame the episode around what Apple Vision Pro signals for builders rather than product impressions. They tee up the core question: what new interactions and startup opportunities emerge when a new computing platform arrives?
- •Episode focus: implications for founders, not headset reviews
- •Vision Pro positioned as a potential new platform shift
- •Open question: what truly takes advantage of spatial computing?
- •Set up Diana as the AR/VR expert voice
- 1:04 – 2:50
Diana’s AR/VR startup story: Escher Reality to Niantic acquisition
Diana explains her YC-backed startup Escher Reality and the original “write once, run anywhere” dream for AR experiences. She recounts why the market took longer than expected and how the company ultimately scaled via acquisition by Niantic.
- •Built an AR SDK for multiplayer AR games across iOS/Android and devices
- •Market timing: AR adoption took longer than early builders expected
- •Outcome: acquisition by Niantic (Pokémon Go)
- •AR platform work shipped to millions of players
- 2:50 – 4:20
Why AR has been hard for decades: optics, field of view, and human vision
The conversation zooms out to the long history of AR/VR attempts and why the problem has resisted mainstream adoption. Diana explains the limitations of optical waveguide approaches and the fundamental challenge of matching the human eye’s field of view and focus behavior.
- •AR/VR has been pursued since the 1960s (Ivan Sutherland)
- •Optical AR (HoloLens/Magic Leap) struggled with narrow field of view
- •Hard problems aren’t just “more compute”—they involve optics/physics
- •Human vision challenges: ~210° FOV and near-infinite focusing ability
- 4:20 – 6:24
Pass-through AR vs optical AR: why Apple’s approach changes the trade-offs
Diana contrasts “see-through optical overlay” AR with video pass-through where the entire view is digitized. She argues pass-through makes some challenges more tractable and opens software tricks that are hard or impossible with purely optical systems.
- •Optical AR overlays pixels onto direct real-world view
- •Pass-through AR turns the world into a high-res video feed (all pixels)
- •Video pipeline enables more flexible rendering and perception tricks
- •Vision Pro aligns more with Quest-style architectures than HoloLens
- 6:24 – 7:41
Eye tracking and foveated rendering: making pixels, heat, and battery work
The group digs into how eye tracking underpins both performance and user experience. Diana explains foveated rendering—high fidelity where you look, lower fidelity in the periphery—as a key trade-off to hit form factor and thermal constraints.
- •Eye tracking is foundational to Vision Pro’s rendering strategy
- •Foveated rendering: sharp at gaze point, blurrier in periphery
- •Trade-offs driven by heat dissipation, battery, and pixel throughput
- •Some perceived blur/artifacts are linked to optics and rendering limits
- 7:41 – 9:24
Hardware vs software: custom silicon, sensors, and ‘self-driving car on your head’
Jared asks how much of Vision Pro’s leap is hardware versus software, and Diana argues it’s both. She highlights the sensor suite and Apple’s R1/M2 architecture, drawing parallels to autonomy stacks that process massive real-time sensor streams.
- •Dual-chip story: M2 for compute + R1 for sensor ingestion/latency
- •Heavy sensor suite: multiple cameras, LiDAR, TrueDepth, inward IR eye cams
- •Real-time processing needs resemble robotics/autonomy constraints
- •Apple leverages iPhone-era expertise in custom silicon and sensing
- 9:24 – 11:06
AR/VR and robotics link: SLAM, localization, and autonomy parallels
Harj prompts the connection to self-driving, and Diana explains SLAM as the shared core problem: locating and mapping in 3D from sensor data. The headset version is harder due to extreme power/thermal limits and wearing comfort.
- •Shared foundation: SLAM (simultaneous localization and mapping)
- •Self-driving uses LiDAR/radar/cameras to understand 3D space
- •Headsets face tighter constraints: weight, heat, and power on a person
- •iPhone/iPad sensor evolution paved the way (Face/Depth, LiDAR)
- 11:06 – 12:45
Why Apple is emphasizing productivity (and why that matters)
Garry argues Vision Pro’s bet is daily utility—work and information—rather than primarily gaming. Diana sees this as the path to “replacing screens,” with eye/hand interactions potentially becoming as foundational as touch was for phones.
- •Apple positioning: productivity-first vs Meta’s gaming-first DNA
- •Potential market: replacing/augmenting many existing screens
- •Natural interaction: gaze + pinch feels quickly learnable
- •Analogy to iPhone HIG: a new interaction paradigm is forming
- 12:45 – 15:26
New interaction primitives: eye tracking as the ‘capacitive touch’ moment
They discuss how Apple’s Human Interface Guidelines and strong eye tracking could catalyze new UX conventions. Like early mobile, third-party developers may invent the signature interaction patterns that become standard.
- •Vision Pro HIG emphasizes depth, space, and gaze-based UX
- •Eye tracking may become a core primitive like touch on iPhone
- •Past skepticism: eye-tracking fatigue was partly due to poor hardware
- •Third-party innovation historically defined major gestures (e.g., pull-to-refresh)
- 15:26 – 16:22
Meta SDK vs visionOS: game-engine roots vs ‘real-world apps’ ergonomics
Jared asks about developer differences between Meta and Apple ecosystems. Diana notes Meta’s strong Unity/Unreal support is great for games, but visionOS can make traditional app tasks (like PDFs) far simpler, reflecting Apple’s app-platform DNA.
- •Meta: strong Unity/Unreal integration from gaming heritage
- •Spatial computing apps often need more than a ‘constrained 3D world’
- •Game engines can be awkward for everyday document/app workflows
- •visionOS lowers friction for standard app primitives (e.g., PDFs)
- 16:22 – 20:18
Is Vision Pro an iPhone moment? Adoption curves and timing for startups
The hosts debate whether this is the beginning of a long arc (like iPhone pre-App Store maturation) or a late-stage inflection. They conclude that big companies often arrive years later, once adoption and platform stability reach critical mass.
- •iPhone parallel: meaningful companies emerged ~5 years after launch
- •Early app ecosystems often start with frivolous use cases
- •Mass-market inflection depends on widespread device penetration
- •Platform fragmentation historically blocked earlier mobile waves
- 20:18 – 23:05
Pre-mortem on Vision Pro: the ‘Tesla Roadster’ risk and ecosystem chicken-and-egg
Jared introduces a pre-mortem: Vision Pro could succeed as a premium niche but fail to reach mass affordability and scale. Diana emphasizes the founder dilemma—whether to bet early without users—and the need for an app ecosystem to justify cheaper successors.
- •Failure mode: remains a high-end ‘Roadster’ without a ‘Model 3’ follow-up
- •Chicken-and-egg: developers need users; users need compelling apps
- •Unlike mobile winners, many apps can’t be cross-platform easily today
- •Near-term wedge may be high-end professional workflows (CAD/engineering)
- 23:05 – 27:58
What founders should build: first-principles thinking, pro use cases, and obsession signals
The panel shifts to practical advice: avoid hype, evaluate ideas from first principles, and look for founders with deep conviction. They highlight professional contexts (dense information work) and note YC’s bias toward builders who are irrationally compelled to work in VR/AR.
- •Potential early markets: high-value info work (e.g., trading/multi-screen setups)
- •We may be in an awkward phase: many apps still feel like flat 2D windows
- •YC approach: fund real value, not cargo-cult platform chasing
- •Best signal: sustained obsession and prior VR building that’s hard to fake
