BlueRun Ventures Leads Round as SiClink Focuses on Visual Reconstruction, Exploring New Frontiers in Brain-Computer Interfaces | BlueRun Portfolio

Invasive brain-computer interface (BCI) startup SiClink has recently completed two consecutive funding rounds — a seed round and an angel round — totaling tens of millions of RMB, with BlueRun Ventures leading the investment and Hillhouse and CAS Shenguang joining as co-investors. The race for next-generation BCIs is shifting from "who can read neural signals more accurately" to "who can build a truly bidirectional, high-speed channel between silicon and carbon."

As applications like motor control and speech decoding gradually become engineered solutions, a more fundamental question emerges: will BCIs remain confined to a handful of medical indications, or will they become foundational infrastructure for the next generation of human-computer interaction?

SiClink is tackling this question from the hardest possible angle: visual reconstruction.

In SiClink's view, what ultimately determines the ceiling of BCIs is no longer whether we can "read the brain," but whether we can establish a high-bandwidth, low-latency information interface between the brain and the external world — one that remains stable for years or even a lifetime.

BlueRun Ventures stated that the endgame of BCI competition is not "whether we can read the brain," but whether we can build a high-bandwidth, long-term stable bidirectional channel between carbon and silicon. By entering through visual reconstruction — the most complex and hardest-to-validate scenario globally — SiClink is subjecting its bidirectional closed-loop capabilities to the most rigorous test. Dr. Fei He has spent over a decade working on neural interface materials, low-threshold stimulation, and high-dynamic-range visual decoding, building systematic technical expertise at the "last 1μm." Moving from unidirectional reading to bidirectional interaction, from medical tool to scalable neural computing platform — BlueRun looks forward to witnessing BCIs become next-generation human-computer interaction infrastructure alongside SiClink.

Over the past few years, the BCI narrative has largely centered on motor control: enabling paralyzed patients to use their thoughts to control robotic arms or cursors. This path validated the feasibility of "reading" neural signals and completed the first wave of translation from research to clinic.

But vision is the perceptual gateway with the highest information density and broadest application boundaries.

Compared to motor control, visual reconstruction is not merely the restoration of sensory function — it represents a leap in how humans and machines exchange information. In the future, vision will no longer be limited to natural perception; it will become a critical interface connecting the physical world, the digital world, and intelligent systems.

A key distinction matters here: it's visual reconstruction, not visual restoration.

Traditional visual restoration is closer to unidirectional encoding — external devices capture images and "write" information into the visual pathway through stimulation. What SiClink emphasizes as visual reconstruction requires both "reading" and "writing," forming a closed read-write loop: the system must first understand what the user is seeing and what visual environment they are in, then write virtual information back into the brain in an appropriate manner.

Bidirectional interaction is the critical inflection point for BCIs to evolve from "functional compensation tools" to "information interfaces": this means visual reconstruction is not limited to any single medical indication. If a system can achieve stable interaction through a pathway as high-bandwidth and complex as vision, the underlying capabilities it develops will have the potential to migrate to broader neural modulation and carbon-based life enhancement applications.

Current BCI technologies either unidirectionally "read" (decoding intent) or unidirectionally "write" (generating perception). But visual reconstruction demands both simultaneously: the system must decode visual information the brain is processing on one hand, tackling challenges like neural representation drift; on the other hand, it must safely and precisely encode new information into the visual cortex with its complex topographic mapping, continuously calibrating within a near-real-time closed loop.

The limitations of existing technical approaches become rapidly amplified once they enter scenarios demanding higher bandwidth, longer lifespan, and stronger closed-loop performance.

Readout systems designed for motor control and unidirectional stimulation systems designed for blind vision restoration cannot naturally migrate into visual reconstruction systems.

SiClink was founded in December 2025. The name "SiClink" evokes "silicon-carbon connection" — using silicon-based technology to connect with carbon-based living systems.

The company's founder, Dr. Fei He, has spent over a decade working at the frontier of invasive BCIs. His research has long focused on the "last 1μm": how to make neural interfaces thinner, softer, and more stable; how to make neural stimulation faster, more precise, and safer; and how to make bidirectional read-write links truly form closed loops.

From 2016 to 2021, he conducted research on ultra-flexible neural interfaces in the labs of Chong Xie and Lan Luan at the University of Texas at Austin and Rice University, achieving multiple milestone results in areas including high-biocompatibility electrode materials (Biomaterials, 2022), low safety-threshold stimulation (Cell Reports, 2023), and high-dynamic-range visual decoding (Nature Communications, 2025). Together, these accomplishments established Dr. He's comprehensive expertise across neural interface materials, bidirectional interaction, and visual reconstruction — providing solid foundational support for SiClink's subsequent technical roadmap.

While the industry remained largely focused on motor control and unidirectional neural modulation, Dr. He was among the earlier voices to propose and consistently build toward visual reconstruction. For him, visual reconstruction is not a single-point application, but a system-level proposition for testing whether BCIs possess the capabilities required for next-generation information interaction.

The global BCI industry is entering a delicate phase. Invasive BCIs have demonstrated the potential to read neural signals through multiple human trials, but the industry still needs to answer a more practical question: will BCIs remain confined to a handful of medical indications, or will they become foundational infrastructure for next-generation human-computer interaction?

SiClink's answer is that starting from visual reconstruction, brain-computer interfaces will move from medicine toward broader applications.

Visual reconstruction is difficult enough and fundamental enough. It demands systems with high-bandwidth neural interfaces, high-density and low-threshold stimulation, long-term and instantaneous stability, bidirectional closed-loop interaction, and intelligent encoding-decoding capabilities. Any weak link in this chain will constrain the final system's usability.

But precisely because of this, once visual reconstruction is engineered and validated, its underlying capabilities are expected to spill over into broader BCI applications.

From unidirectional reading and writing, to bidirectional interaction. From medical device, to information interface. From single-function validation, to scalable neural computing platform.

The next decade of brain-computer interfaces may well begin here.

BlueRun Ventures Leads Investment as Naosic Raises Over $10 Million in Angel Round, Focusing on Invasive BCI and Neural Foundation Models | BlueRun Portfolio

BlueRun Early-Stage Portfolio Company Tian Cheng Technology Raises Another 100-Million-RMB B+ Round, Accelerating Exoskeleton "Human-Machine Fusion" New Consumer Era | BlueRun Portfolio

BlueRun Angel Portfolio Company Kaiwu Ji Completes 100-Million-RMB Angel+ Round, AI4S Driving Construction of Materials Intelligence Infrastructure | BlueRun Portfolio