Gasgoo Munich- Physical AI company SynapX announced it has secured nearly $50 million in its maiden funding round, with backing from Horizon Robotics, GL Ventures, Xiaomi Strategic Investment, ShunWei Capital, and Linear Capital.
Founded in January 2026, SynapX is dedicated to building sustainable, self-evolving embodied intelligence. Its founder, Du Dalong, previously established PhiGent Robotics and was the No. 6 founding employee at Horizon Robotics, as well as a founding member of Baidu's IDL. Co-founders Liang Zhujin, Pan Yangjiayi, and Fan Qingyuan also bring deep experience from Horizon or PhiGent Robotics.
Image Source: SynapX
The move from inception to closing its first round took SynapX less than two months — a rapid pace that underscores the market's confidence in its founding team, technical roadmap, and long-term value.
Proceeds from this round will primarily fuel core technology and product development, data infrastructure construction, and talent acquisition. Meanwhile, the company is already advancing its next financing round, drawing continued interest and support from global industrial capital, family offices, and top-tier US dollar funds.
Alongside the funding news, SynapX unveiled its SYNTH architecture. In the company's view, true embodied intelligence goes beyond understanding the world; it requires executing operational tasks within real physical environments and forming a closed execution loop. Achieving this depends not just on model scale, but on the unified design of model architecture, world modeling, and data systems.
Image Source: SynapX
Specifically, the SYNTH architecture comprises SYNAction, a model architecture for operational intelligence; SYNWorld, an "omniscient" foundation model for understanding the physical world; and SYNData, a multimodal data system driving continuous model evolution.
For SYNAction, SynapX pioneered REMA, a split-frequency, multi-scale, end-to-end operational architecture. Within a unified model, it achieves a closed loop from cognition to execution across three layers: System 2 (low frequency) handles task understanding, cognitive reasoning, and high-level planning; System 1 (mid frequency) generates action strategies, bridging cognition and execution; and System 0 (high frequency) manages underlying control and fine-grained interaction.
In the SYNWorld component, SynapX developed VFT-WFM, a framework that unifies vision, force, and touch into physical interaction modeling. This enables robots to not only recognize what they see but also understand how to execute operations, make contact, and apply force.
For SYNData, SynapX constructed OPDS and introduced the concept of "AI as a Sensor." Leveraging unique hardware designs and innovative AI algorithms, the system enables high-precision collection and scalable generation of multimodal physical signals — including vision, force, and touch — providing the high-quality data needed for continuous model evolution.
About "Seeds Discovery":
Gasgoo's "Seeds Discovery" column is designed to build a platform connecting startups, industrial ecosystem partners, investors, and local governments, empowering the upstream and downstream supply chain. Since its launch, the column has focused on identifying standout companies, technologies, and business models that offer inspiration and leadership during the wave of intelligent transformation. According to Gasgoo, nearly every startup covered by "Seeds Discovery" has successfully connected with industrial ecosystem resources.
