News

Startups

Artificial Intelligence

Asia

The Invention Lab and QRT Back Singapore AI Startup RIDM in Seed Round

The Invention Lab has finalized a seed investment in RIDM, a Singaporean deep-tech startup tackling the critical Memory Wall bottleneck in AI computing.

The Invention Lab has finalized a seed investment in RIDM, a Singaporean deep-tech startup tackling the critical Memory Wall bottleneck in AI computing.

NewDecoded

Published Feb 28, 2026

Feb 28, 2026

3 min read

The Invention Lab announced on February 27 the completion of a seed investment into RIDM PTE. LTD., a Singapore-based startup focusing on next-generation AI architecture. This round was joined by London-based quantitative investment firm Qube Research & Technologies (QRT), signaling strong institutional interest in the company's technical foundations. The deal marks a significant step for the deep-tech firm as it moves toward full commercialization of its hardware solutions.

RIDM is developing a proprietary system called Dynamically Orchestrated Dataflow Architecture, or DODA. This technology is designed to overcome the Memory Wall, a persistent hardware limitation where data transfer speeds cannot keep up with processor requirements in high-performance computing environments. By restructuring how data flows during processing, the company aims to provide far greater efficiency than standard architectures.

Unlike traditional setups that require complex hardware languages, RIDM allows developers to use standard C++. Their breakthrough involves solving spatial compilation in linear time, which significantly simplifies the process of mapping software to specialized hardware like FPGAs. This approach lowers the barrier for engineers and offers a competitive development environment for high-performance tasks.

The company is a spinoff from the National University of Singapore (NUS) and is led by CEO Jin-ho Lee, who holds a doctorate in dataflow architecture. Technical advice is provided by Associate Professor Trevor E. Carlson, the lead developer of the renowned SniperSim multi-core simulator. This combination of academic expertise and industry experience has positioned RIDM as a serious contender in the semiconductor IP space.

Currently, RIDM is collaborating with QRT on a proof of concept to verify the efficiency and low latency of the DODA architecture. The startup holds exclusive licenses to three international patents and plans to leverage these results to pursue further global strategic partnerships. CEO Jin-young Kim of The Invention Lab noted that the investment represents a bet on a team capable of breaking through the limitations of traditional computing.


Decoded Take

Decoded Take

Decoded Take

This investment highlights a growing shift away from the traditional von Neumann architecture that dominates today's computing landscape. By backing a startup that simplifies FPGA programming through a general-purpose language like C++, investors are betting on a more accessible alternative to the NVIDIA CUDA ecosystem. The partnership with a quantitative firm like QRT is particularly telling, as it suggests that the immediate commercial value of RIDM's tech lies in sectors where every microsecond of latency counts. As AI models continue to outpace hardware capabilities, specialized dataflow architectures like DODA may become the necessary standard for high-speed financial and research applications.

Share this article

Related Articles