MountAIn raises €7.3 million to bring AI applications to chips costing under $10
The Grenoble startup says MountainCore adapts entire vision applications—not just models—for small processors. The financing backs commercialization and international growth.
Rather than building silicon, MountAIn is using its €7.3 million seed round to scale MountainCore, software that ports complete computer-vision applications—not just models—to sub-$10 microcontrollers. The company says the approach can cut deployment work from months to hours, reduce model memory by up to threefold and preserve accuracy without retraining; these performance claims have not been independently verified. Funding will expand its Grenoble engineering team and support growth in Europe and Silicon Valley, with CES 2027 as its next showcase.
01
The November 17, 2026 round was backed by 360 Capital, Supernova Invest and Crédit Agricole Alpes Développement, with additional support from Bpifrance.
02
MountainCore supports multiple ARM microcontroller families, aiming to reduce hardware-specific optimization work.
03
MountAIn cites demonstrations with Teledyne e2v and partnerships with Roboflow and Alif Semiconductor as evidence of industry engagement.
MountAIn has secured €7.3 million to commercialize software aimed at putting computer vision inside inexpensive devices. The Grenoble company announced the seed financing on November 17, 2026. Its pitch is to adapt entire applications for chips costing less than $10, reducing the specialized engineering needed to move AI into cameras and industrial equipment.
360 Capital, Supernova Invest and Crédit Agricole Alpes Développement backed the financing, with additional support from Bpifrance. The announcement places MountainCore, MountAIn’s software platform, at the center of its commercial growth plans rather than presenting the round as the launch of a new processor.
Adapt the application, not just the model
MountAIn frames the problem as a choice manufacturers face when small processors cannot handle sophisticated AI applications. They can buy more powerful, energy-hungry hardware or send data to the cloud. The latter brings connectivity requirements, latency, expense and privacy concerns. Adapting software to smaller chips can itself require months of specialist work.
MountainCore takes a software route around those constraints. The company says it automatically adapts complete computer vision applications, including the AI model and its supporting software, for microcontrollers: small processors embedded in equipment. The distinction is important to its pitch. Making the model fit is only part of getting the application running on a constrained device.
The company claims MountainCore reduces model memory requirements by up to threefold, uses on-chip AI accelerators more effectively and preserves accuracy without retraining. It supports multiple families of ARM-based microcontrollers, which MountAIn says reduces hardware-specific optimization work. These are company-supplied capability claims, not independently verified performance results.
Industrial demonstrations and an off-grid sensor
The commercial case includes existing demonstrations and development relationships. MountAIn points to work across industrial imaging, AI development tools and semiconductor hardware as evidence that the platform is moving beyond an internal demonstration. Its announcement identifies three connections, each serving a different part of the path from developing a model to running it inside equipment:
Teledyne e2v demonstrated MountainCore at VISION Stuttgart in 2026, combining 2D and 3D vision from its Perciva 5D camera.
Roboflow offers direct model export to MountAIn’s autonomous IBEX vision sensor, according to the announcement.
Alif Semiconductor is partnering with MountAIn on computer vision applications for its Ensemble and Balletto processor families.
IBEX illustrates a different deployment setting from a connected factory. MountAIn developed the sensor to demonstrate MountainCore in remote locations. It says IBEX consumes less than 100 milliwatts, uses solar power and satellite connectivity, and analyzes images locally. Intended applications include early wildfire detection and infrastructure monitoring in places without conventional computing infrastructure.
For deployed applications, MountAIn emphasizes keeping processing on the device rather than sending sensitive data to the cloud. That is the practical link between its two promises: fitting applications onto inexpensive hardware and allowing them to operate locally, with lower cloud costs and data kept at the point of collection.
The funding will support a larger engineering team in Grenoble and a stronger presence in Europe and Silicon Valley, targeting semiconductor, smart-camera and industrial-equipment manufacturers. MountAIn’s next announced showcase is CES in Las Vegas in January 2027, where it plans to demonstrate MountainCore’s industrial capabilities and IBEX applications for remote locations.
Sources
globenewswire.comMountAIn Secures €7.3 Million in Seed Funding to Run Advanced AI on Chips Costing Less Than $10
Reader comments
Newest comments first. Replies stay oldest first.