Meta's reveal of Project Orion at Meta Connect represents something rare in tech: a company showing working hardware that has no commercial timeline and costs roughly three times what a high-end computer costs to manufacture. At $10,000 per unit in production, these aren't prototypes destined for launch in three to five years. They're proof of concept—and evidence of a strategic desperation.
The device itself is technically clever. The glasses use silicon carbide lenses (a material borrowed from semiconductor manufacturing and EV batteries, rarely synthesized on Earth) to project micro-LED displays at angles that glass cannot achieve without color distortion. A wrist-worn EMG band reads muscle signals in the forearm to detect finger gestures, eliminating the need to keep hands visible to cameras. A separate "Puck"—effectively a wireless compute module the size of a power bank—handles processing over a Wi-Fi connection, keeping the glasses themselves light enough to wear all day. Five cameras provide spatial reference, eye-tracking enables gaze-based cursors, and ten custom chips process the infinitely more complex task of placing persistent holograms over live reality rather than rendering a closed virtual environment.
But the reveal exposes a mismatch between engineering ambition and market reality. Meta has spent a decade and roughly $16 billion annually (with no revenue return from Reality Labs) chasing a platform replacement narrative. The company has no choice: it's been subordinate to Apple and Google's platforms for years, asking permission to do what its own data harvesting models require. Showing Orion is an argument to shareholders that this bet matters. Yet the same company struggles with a more immediate problem: these glasses don't replace smartphones because input still works better on a glass rectangle in your hand. Voice and gesture are imprecise for text. Voice in public remains socially awkward. And for markets outside the English-speaking world—Portugal included—voice interfaces in underfunded languages like European Portuguese will lag years behind, if they arrive at all.
Apple, meanwhile, owns supply chains, retail presence, and the ability to launch as a premium device first (like the Vision Pro's $3,500 entry point) before descending in price. Meta owns neither. It licenses Ray-Ban frames from EssilorLuxottica and depends on third-party distribution for Quest headsets, giving it neither the control nor the margin structure to dominate a new compute layer. The more plausible future, one that plays to existing strength, is spectacles as a wearable accessory to the phone—a companion device, not a replacement. Apple and Google could integrate such a thing more naturally into ecosystems they already control. Meta gets squeezed again.
The real cost here isn't silicon carbide or custom chips. It's the years lost while smartphone makers were establishing dominance in AI infrastructure. By the time Orion reaches $3,000–$4,000 (the price Zuckerberg has mentioned), language models and multimodal AI will be mature, cheap, and standardized. Apple's Siri will work in Portuguese. Google's Assistant will handle context across devices. The barrier to entry drops precisely as Meta's advantage erodes.