Kyber, a Paris-based startup building real-time device control infrastructure for robots, drones, and remote systems, has raised $5 million in a seed round led by Lightspeed Venture Partners, with OVNI Capital and Kima Ventures participating alongside the lead, a round that amounts to approximately €4.3 million at current exchange rates.
The company was founded by Jean-Baptiste Kempf, who built VLC, the ubiquitous open-source media player, before serving as CTO at cloud gaming startup Shadow. That biography is not incidental to what Kyber does.
The Technology Behind Kyber’s Real-Time Device Control
Kyber’s core product is an SDK that synchronises video, audio, sensor data, and control inputs across remote devices with minimal delay. According to Startup.eu, the platform achieves end-to-end latency as low as 8 milliseconds and can be deployed on-premise, in the cloud, or in hybrid environments.
That 8ms figure comes from a specific setup: in a February 2025 interview, Kempf described reaching that threshold in a 240Hz glass-to-glass environment, sending video as datagrams rather than reliability-guaranteed streams to keep the pipeline lean. The choice of datagrams over standard streaming protocols is deliberate: dropping a frame occasionally is a tolerable trade-off; adding latency is not.
The architecture reflects Kempf’s streaming heritage directly. The SDK is built on top of FFmpeg, rewired to operate in push mode, and a modified version of VLC running in real-time mode, using the QUIC protocol for transport. The platform supports H.264, HEVC, VP9, and AV1 codecs across Windows, Linux, macOS, iOS, and Android on both client and server sides.
‘If you control things in the real world, every millisecond matters,’ Kempf told TechCrunch. The startup’s name is a nod to the lightsaber crystals in Star Wars, where speed of response is the point.
Robotics, Drones, and the Scale Problem
Lightspeed, which manages over $50 billion in assets and has backed Anthropic and Mistral AI, framed the investment around a simple thesis: ‘Physical AI is only as good as the underlying systems running it.’
The framing suits Kyber’s current focus. The company is prioritising three segments: robotics, drones of every kind, and remote IT access. It is already in commercial deployment across defence, telco, robotics, and AI customers.
Kempf argues the scale opportunity separates Kyber from proprietary solutions competitors have already built internally. ‘But the largest fleets today have maybe 2,000 or 3,000 vehicles,’ he said. ‘Imagine you need to manage millions of them; that’s not the same thing.’
That jump in scale also changes what observability means. When AI agents rather than people are managing entire fleets, knowing that systems are actually functioning becomes structurally more complex. Even at a small scale, there is a practical argument: pushing a software update remotely to a distributed device fleet avoids the cost and friction of physical access entirely.
Kempf’s framing of the addressable market is deliberately broad: ‘all the use cases where the person who’s operating is not in the same place as the compute, which is not in the same place as the action.’ Remote driving, autonomous delivery, distributed industrial sensors, the list is long.
Open Source Roots, Enterprise Revenue
True to Kempf’s history with VLC, the core Kyber project is open source. The company monetises through a productised enterprise version and through forward-deployed engineers who handle custom deployment for clients directly, a model Palantir made familiar in defence and intelligence.
Those forward-deployed engineers make up a sizeable share of Kyber’s current 25-person team. The company is headquartered in Paris, with offices in San Francisco and Singapore to serve what it expects will be a global client base.
In remote IT access, Kempf has positioned Kyber as something more ambitious than a Citrix replacement, though even that comparison points to a substantial market. The company’s careers page distils the logic plainly: ‘The companies that tried to solve it spent years and tens of millions building custom solutions they’ll never share. We’re building the version everyone else can use.’
Whether enterprise deployments in robotics and defence convert fast enough to fund the open-source distribution model is the real test ahead. The first proof will come from whether Kyber’s commercial pipeline grows in proportion to the fleet sizes Kempf is already describing.
