Frequently
Asked Questions

VIM™ is Vertical Compute™ ‘s proprietary technology that combines memory and compute into a single, cohesive solution. It uniquely combines the best aspects of different memory solutions into a single high-density, vertical magnetic memory solution. As such, this new technology will unlock the next wave in AI applications, data processing and beyond by bringing vast amound of data in direct connection to the processing cores.

VIM™ technology is designed to solve the memory bottleneck in modern computing by combining the best attributes of different memory technologies into one. This approach will break the historical trade-offs between performance, density, and cost that have constrained computing for decades. VIM™ is designed to achieve the high speed (<10 ns latency) and unlimited endurance of on-chip SRAM, while also providing the high density and cost-efficiency of 3D-NAND. This revolutionary combination will eliminate the need for a complex and inefficient memory hierarchy. Furthermore, VIM™ ‘s vertical architecture allows to reduce the physical distance between memory and processing units, which is expected to dramatically reduce the power consumption needed for data transfers, improving the overall system efficiency and reducing the energy footprint of AI.

Vertical Compute™ 's VIM™ technology addresses the "memory wall bottleneck" which is hindering the growth of modern AI and high-performance computing. The memory wall bottleneck is caused by the original limitations of the existing memory hierarchy, which relies on a mix of technologies like SRAM, DRAM, and NAND, forcing difficult trade-offs between speed, density, and cost. This exponential growth of AI models has exposed this inefficiency, leading to a significant performance bottleneck that cannot be solved with incremental improvements of existing memory solutions. Vertical Compute™ solves this problem with its proprietary VIM™ technology, a new memory solution that combines the high speed and endurance of SRAM with the high density and cost-efficiency of 3D-NAND into a single, on-chip solution. This approach eliminates the need for a complex memory hierarchy, enabling AI models to run directly on local devices.

VIM™ technology will unlock a new generation of AI systems for both training and inference workloads by solving the memory wall bottleneck. More precisely, Vertical Compute™ ‘s technology will enable the use of a high-capacity, high-speed memory for large language models (LLMs) and other AI applications, both in data centers and on local devices. With its ability to provide high-performance, non-volatile memory on a single chip, VIM™ becomes ideal for edge computing applications that require real-time processing and low latency, such as autonomous vehicles, smart devices, and IoT sensors. Hence, VIM™ enables future devices to run complex AI models without relying on a constant network connection to the cloud. Next, VIM™ can also serve as a powerful alternative to HBM in High-Performance Computing (HPC) systems by providing higher performance at a much lower cost. Its ability to handle massive data throughput makes it a prime candidate for scientific simulations, data analytics, and other compute-intensive workloads. Finally, as AI becomes more integrated into everyone’s daily life, VIM™ can enable a new generation of consumer devices, such as laptops and smartphones, with powerful, on-chip AI assistants that are both more responsive and more private.

Vertical Compute has secured €20 million in a seed investment round that was led by imec.xpand, with key participation from XAnge, Eurazeo, Vector Gestion, and imec.

Vertical Compute will use its funding to develop their VIM™ technology and assemble a world-class team of engineers to accelerate its R&D roadmap.