Analysis
BlackBerry, a company most people associate with a keyboard phone that stopped mattering in 2013, is making its case as an infrastructure supplier to the physical AI buildout. The asset is QNX, the real-time operating system it acquired in 2010 and which now sits inside automotive digital cockpits, advanced driver assistance systems and industrial controllers. CNBC reported on the company's push into cars and robots.
The pivot has been long and expensive. BlackBerry stopped making its own handsets in 2016 and reinvented itself as a software and security company, an identity it never fully settled -- it sold its Cylance endpoint security business to Arctic Wolf in 2025 to concentrate on the embedded and secure communications businesses. QNX is the piece that survived every strategic reversal, because it does something genuinely hard: deterministic, safety-certified execution that automakers can put through ISO 26262 certification.
The competitive picture
QNX is not unopposed. Automotive Grade Linux and various in-house Linux builds are cheaper, and Google's Android Automotive has taken the infotainment layer at Volvo, Polestar, GM and Ford. But infotainment and safety-critical control are different problems, and certified real-time systems remain a small club: QNX, Wind River's VxWorks, Green Hills' INTEGRITY. Robotics widens that market -- humanoid and industrial robot makers face the same certification requirements automakers do, and most are not going to write a real-time kernel themselves.
What is unproven
A credible technical position is not the same as a growth business. BlackBerry's revenue has been roughly flat for years, and design wins in automotive convert to royalty revenue on multi-year lags tied to vehicle production volumes -- which have themselves been soft. Robotics is a real option, but robot unit volumes today are a rounding error against auto production.
The financial shape of the business explains the market's skepticism. QNX earns development seat revenue up front and royalties per shipped unit, which means today's design wins convert to revenue three to five years out as vehicle programs reach production. That lag makes the business genuinely predictable and genuinely slow, and it leaves BlackBerry exposed to automaker production volumes it does not control.
Robotics is the more interesting option because the certification requirement is identical and the incumbent answer is worse. Most robotics companies today run ROS 2 on Linux, which is excellent for research and not certifiable for safety-critical control -- a gap that becomes acute the moment a robot operates near people in a regulated workplace. QNX has spent three decades passing exactly those audits. Whether that converts depends on whether humanoid and industrial robot volumes materialize at a scale worth the integration cost, and today they do not.
Investors have heard versions of this pitch before, which is the honest problem. BlackBerry has announced strategic refocusings repeatedly since 2013, and each one was accompanied by a credible-sounding asset story. The difference this time is that the asset in question is already shipping inside production vehicles rather than being pitched as a future platform.
The measurable signal is royalty revenue from the QNX design-win backlog, and whether any named humanoid or industrial robotics program discloses QNX as its runtime.
