Jean-Baptiste Kempf, the visionary behind the ubiquitous VLC Media Player, is shifting his focus from ensuring your videos play smoothly to making robots operate seamlessly in the real world. His new venture, Kyber, aims to provide the foundational infrastructure for what he believes will soon be hundreds of millions of robots and drones. This move signifies a critical evolution in how we interact with and deploy physical AI.
What Happened
Kempf, an open-source pioneer, has launched Kyber as an infrastructure layer designed to control remote devices in real-time. The company's core offering is an SDK that meticulously synchronizes video, audio, sensor data, and control inputs, prioritizing minimal latency. This capability is paramount for physical AI applications, where delays can have significant operational consequences.
Kyber's technical approach is deeply rooted in video-streaming technology, a natural extension of Kempf's background at VLC and his previous role as CTO at cloud gaming startup Shadow. The platform, however, also integrates crucial IoT expertise, focusing on optimizing performance across a range of devices and at immense scale. The company has raised $5 million in funding, led by Lightspeed, an investor known for backing AI giants like Anthropic and Mistral AI, underscoring the strategic importance of robust underlying systems for physical AI.
While the core project remains open source, true to Kempf's ethos, Kyber also offers a productized version for enterprise customers. This includes hands-on custom deployment services provided by "forward-deployed engineers" (FDEs) – a model similar to companies like Palantir. This hybrid approach aims to cater to both the broader developer community and large organizations requiring tailored, scalable solutions for managing fleets that could eventually number in the millions.
Why It Matters
For developers and IT professionals, Kyber addresses several fundamental challenges posed by the proliferation of physical AI and remote-controlled systems:
- Low-Latency Control for Physical AI: The efficiency of AI agents in the physical world is directly tied to the speed and reliability of their control systems. Kyber's focus on minimizing latency for synchronized data streams (video, audio, sensors, control) is crucial for applications like autonomous vehicles, remote surgery, industrial robotics, and drone operations. Every millisecond matters when physical actions are involved, directly impacting safety, precision, and operational effectiveness.
- Scalability for Distributed Fleets: Current large-scale robot fleets number in the low thousands. Kempf's vision of managing "millions" of devices presents a formidable challenge in terms of infrastructure, networking, and data management. Kyber's ambition to provide a unified platform for such scale implies sophisticated solutions for device provisioning, secure communication, over-the-air updates, and distributed command and control. This capability is essential for enterprises looking to deploy vast networks of robotic systems.
- Enhanced Observability and Management: As operations scale and AI agents take on more autonomous roles, comprehensive observability becomes non-negotiable. Knowing that systems are working correctly, diagnosing issues remotely, and pushing software updates without physical access are vital for maintaining uptime and reducing operational costs. Kyber's platform aims to provide the necessary tools to monitor and manage these complex, distributed environments efficiently.
- Bridging Open Source and Enterprise Needs: By offering an open-source core alongside enterprise-grade support and services, Kyber creates a flexible ecosystem. Developers can leverage the open-source SDK for innovation and custom projects, while enterprises gain the reliability, tailored solutions, and dedicated engineering support required for mission-critical deployments. This model can accelerate adoption and foster a community around physical AI infrastructure.
What To Watch
Kyber's trajectory will be a fascinating one to follow. Key areas to observe include:
- Open-Source Adoption: How quickly will the developer community embrace Kyber's open-source SDK? Its success will depend heavily on ease of integration, performance benchmarks, and the richness of its feature set for diverse robotic applications.
- Enterprise Deployments: The real test of Kyber's scalability and robustness will come with its initial large-scale enterprise deployments. Observing how Kyber tackles the complexities of managing tens or hundreds of thousands of devices in real-world industrial, logistical, or urban environments will be critical.
- Evolution of Physical AI: Kyber is positioned to be a foundational layer for the physical AI revolution. Its growth will likely mirror advancements in robotics, autonomous systems, and edge computing. The capabilities Kyber enables could unlock new categories of applications previously hindered by latency or scalability constraints.
Jean-Baptiste Kempf's pivot from VLC to robotics underscores a significant trend: the underlying infrastructure for real-time data and control is becoming as crucial for physical machines as it has been for digital media. Kyber's success could be a major step towards making the widespread deployment of intelligent robots a smooth reality.