Lina Hope
If you have ever opened a video file on a computer, you have likely encountered the orange traffic cone icon of VLC Media Player. With over 6 billion downloads globally, it is arguably the most recognizable piece of open-source software in history. Its lead developer, Jean-Baptiste Kempf, is now betting that the next wave of ubiquitous technology won’t just live on your screen—it will be roaming your streets.
Kempf is the architect behind Kyber, a Paris-based startup building what he describes as the "infrastructure layer" for the era of physical AI. As robots, drones, and autonomous machines move from factory floors into the public sphere, the challenge of controlling them in real-time has become a critical bottleneck. Kyber is attempting to solve this by creating a universal framework for remote device management, a move that has recently attracted a $5 million seed round led by Lightspeed Venture Partners.
Main Facts: Solving the Latency Problem
At its core, Kyber is an SDK (Software Development Kit) designed to synchronize video, audio, sensor data, and control inputs across distributed networks with near-zero latency. The company’s value proposition is simple yet profound: in a world where machines are increasingly autonomous, there must be a seamless, reliable, and ultra-fast way to bridge the gap between the human operator (or the remote AI brain), the compute cloud, and the physical action.
Kempf’s vision is rooted in a fundamental reality of robotics: every millisecond matters. Whether a drone is navigating a narrow corridor or a remote-controlled vehicle is responding to a traffic hazard, the lag between a command and an action can mean the difference between a successful operation and a catastrophic failure. By leveraging high-performance streaming technology—honed during Kempf’s tenure as CTO of the cloud gaming startup Shadow—Kyber ensures that the “eyes” and “hands” of these machines remain perfectly in sync with the remote intelligence controlling them.
Chronology: From Media Players to Machine Intelligence
The genesis of Kyber can be traced back to the convergence of Kempf’s open-source pedigree and the rapid maturation of the physical AI sector.
- The VLC Era: Kempf became a legendary figure in the open-source community through his leadership at VideoLAN, the nonprofit behind VLC. This period cemented his expertise in managing complex, cross-platform video streaming at massive scale.
- The Cloud Gaming Pivot: While serving as CTO of Shadow, a company pushing the boundaries of cloud-based gaming, Kempf faced the harsh realities of remote interaction. He realized that the same technologies used to stream high-fidelity games with minimal lag could be repurposed for the much more critical world of robotics and remote industrial hardware.
- The Birth of Kyber: Drawing inspiration from the "Kyber crystals" of Star Wars—which provide the power for lightsabers—Kempf founded the company to create the foundational "crystal" for robotic connectivity.
- The Funding Milestone: In late 2024, the company secured $5 million in seed funding from Lightspeed. This investment signaled a broader trend in venture capital: a shift away from purely digital AI toward "Physical AI," or systems that can interact with the tangible world.
Supporting Data: Why Scale is the Ultimate Challenge
While many companies have built proprietary remote-control solutions for specialized use cases, these systems rarely scale beyond a few hundred units. Kempf argues that the world is heading toward a future populated by millions of autonomous drones and robots, a paradigm shift that legacy systems are ill-equipped to handle.
"The largest fleets today have maybe 2,000 or 3,000 vehicles," Kempf noted. "Imagine you need to manage millions of them; that’s not the same thing."
To manage a fleet of that magnitude, the infrastructure must handle several layers of complexity:
- Observability: Operators need to know that millions of devices are functioning correctly in real-time.
- Resource Optimization: The software must tune itself to the specific compute power available on each device, whether it is a high-powered autonomous car or a low-energy, battery-constrained drone.
- Deployment: Kyber’s model includes a hybrid approach, offering an open-source core for developers alongside an enterprise version that provides custom, hands-on support through "Forward-Deployed Engineers" (FDEs).
The current team of 25, headquartered in Paris with outposts in San Francisco and Singapore, is already deploying these solutions across defense, telecommunications, and robotics sectors, validating the market’s hunger for a standardized, scalable protocol.
Official Responses and Strategic Vision
The investment from Lightspeed Venture Partners was accompanied by a clear thesis regarding the future of the AI stack. "Physical AI is only as good as the underlying systems running it," the firm stated in a recent LinkedIn announcement. By backing Kyber, Lightspeed is essentially betting on the "plumbing" of the robotics revolution rather than the robots themselves.
Kempf’s approach to the market is refreshingly utilitarian. When asked about potential competitors, he points to the fragmented landscape of custom, closed-source solutions. "The companies that tried to solve it spent years and tens of millions building custom solutions they’ll never share," he remarked. "We’re building the version everyone else can use."
This philosophy mirrors the VLC model: make the base technology so accessible and reliable that it becomes the industry standard by default. By offering an open-source foundation, Kyber encourages widespread adoption, while the enterprise-grade version—complete with FDE support—secures the revenue stream necessary to maintain the ecosystem.
Implications: The Future of Remote IT and Beyond
The implications of Kyber’s technology extend far beyond the flashy headlines of autonomous drones. One of the most significant, if less "glamorous," applications is remote IT access.
As businesses transition to more decentralized, robotic-heavy workflows, the ability to maintain these systems without physically touching them becomes a core operational requirement. Kyber aspires to challenge incumbents like Citrix, not by merely offering a remote desktop, but by providing a high-speed, low-latency conduit that treats a robot in a warehouse or a sensor in a remote power grid as if it were sitting on a local desk.
The Rise of the Agentic Fleet
As AI agents move from chatbots to "do-bots"—systems that manage physical networks and fleets—the requirement for a robust communication layer will become non-negotiable. If an AI agent is in charge of a fleet of delivery drones, it must have a high-fidelity, low-latency connection to every device to manage maintenance, software updates, and emergency override protocols.
Kyber’s focus on this "infrastructure layer" positions it at the intersection of three major trends:
- The Edge Computing Revolution: Shifting intelligence closer to the source of data.
- The Democratization of Robotics: Lowering the barrier for companies to deploy autonomous fleets.
- The Open-Source Standard: Creating a common language for device interaction that prevents vendor lock-in.
Conclusion
Jean-Baptiste Kempf’s transition from the world of media players to the world of robotics is more than just a career pivot; it is a recognition that the digital and physical worlds are rapidly merging. Just as VLC standardized how we consume media across a fragmented digital landscape, Kyber aims to standardize how we control the physical machines that will define the next decade of industry.
Whether Kyber will achieve the same ubiquity as the orange traffic cone remains to be seen. However, by solving the fundamental, unglamorous problems of latency, scalability, and observability, the company has positioned itself as an essential utility for the future of physical AI. In a market often obsessed with the "AI brain," Kyber is a timely reminder that without a high-speed, reliable nervous system, even the smartest robots are destined to stay grounded.
Asep Darmawan
Muslim
