Former Building 8 employees shed light on Facebook's challenges in trying to break into consumer hardware, as it struggles to turn Portal into a success
KEY POINTS — Building 8, the laboratory she started, turned into a costly failure for Facebook, and Dugan was gone in less than two years.
Context & Ripple Effects
The arc runs from promise to quiet dismantlement: Facebook hired Regina Dugan away from Google's ATAP in 2016 with a budget of hundreds of millions of dollars to run Building 8 (hired her as its new R&D lead), and by early 2017 the group was touting at least four experimental hardware projects with a showcase slot at F8. Within eighteen months she was gone (announced her departure), and by late 2018 the lab itself had been dissolved — renamed Portal and stripped of its more experimental work into Facebook Reality Labs (renamed Building 8 to Portal).
This CNBC piece adds the insider post-mortem: former employees describing a lab that burned cash without shipping, compounded by an earlier-reported failure to build Portal's own voice recognition amid chaotic staff rotation and confused priorities (failed to develop its own voice recognition for Portal). The through-line is that Facebook's consumer hardware effort now rests entirely on Portal, a product still searching for traction.
First-order effects
- Facebook's hardware ambitions have collapsed from a multi-project moonshot lab under a marquee hire to a single product line — Portal — that the company is still struggling to make successful.
- The experimental pipeline that was supposed to feed future devices now lives inside Facebook Reality Labs, which inherited Building 8's more speculative projects when the lab was renamed.
Second-order effects
- A public accounting of staff rotation and confused priorities raises the cost of Facebook's next attempt to recruit star hardware leadership — candidates can read exactly how the last bet on a DARPA-style lab ended.
- Portal's dependence on outside voice technology, after the in-house effort failed, leaves the device's core experience tied to components Facebook never controlled.
Third-order effects
- If the pattern holds, big software firms learn that importing a star executive and a skunkworks structure does not buy hardware competence — the durable capability sits in accumulated engineering teams, not lab charters.
- Consumer-hardware efforts at ad-driven platforms face a structural test: without a services or commerce rationale that justifies years of losses, labs get folded into nearer-term divisions like Reality Labs rather than shut down outright.
The trend: Platform companies' attempts to buy hardware credibility through high-profile lab hires are giving way to consolidation around single products and existing divisions, with talent retention emerging as the binding constraint.