Vertical Compute develops vertically integrated memory chiplet technology for AI computing, targeting energy savings and performance gains by reducing the memory bottleneck in AI hardware systems.
Vertical Compute, founded in 2024 and headquartered in Louvain-La-Neuve, Belgium, is developing vertically integrated memory technology aimed at addressing one of the most persistent bottlenecks in AI computing: the gap between processing speed and memory access. The company operates across multiple sites in Belgium and France, including Leuven, Grenoble, Nice, and Paris, positioning itself within Europe's established semiconductor and microelectronics corridors.
The company's core product, VIM™ (Vertical Integrated Memory), uses a chiplet-based architecture to integrate memory and compute more closely together. According to the company, this approach targets latency figures of 10 nanoseconds and sub-picojoule-per-bit efficiency, with claims of up to 80% energy savings and 100X performance gains over conventional memory configurations in AI workloads. The technology sits at the intersection of semiconductor design, memory technology, and AI computing - a domain attracting significant interest as large-scale AI models increasingly strain existing hardware architectures.
Vertical Compute raised €20 million in seed funding. The company has signalled a focus on building what it describes as an elite technical team, with an emphasis on European semiconductor innovation and the development of sustainable, affordable AI infrastructure. No further details on team size or commercial deployment timelines have been disclosed.
