🔥36氪•Stalecollected in 10h
Peking University Scientists Launch BCI Startup
#bci#neuroscience#embodied-ai#hardwarexin-sheng-shi-jie-(brain-computer-interface)peking universityxin sheng shi jieneuralink
💡Learn how top scientists are building 'neural graphics cards' to enable high-bandwidth BCI for visual reconstruction.
⚡ 30-Second TL;DR
What Changed
The startup focuses on high-bandwidth, integrated 'neural graphics card' systems for visual and language reconstruction.
Why It Matters
This development highlights the growing intersection of BCI and embodied AI, pushing the boundaries of human-machine interaction through high-throughput neural data processing.
What To Do Next
Follow the development of high-throughput neural decoding algorithms, as they will be critical for future human-AI integration interfaces.
Who should care:Researchers & Academics
Key Points
- •The startup focuses on high-bandwidth, integrated 'neural graphics card' systems for visual and language reconstruction.
- •The team includes experts in neuroscience, BCI hardware, and embodied AI, aiming to bridge silicon-based computing with biological neural systems.
- •The company has successfully taped out 28nm BCI chips and is developing higher-throughput, bidirectional interaction systems.
🧠 Deep Insight
Background and context from public sources — not the original article. 10 sources cited.
🔑 Enhanced Key Takeaways
- •Xin Sheng Shi Jie, also referred to as Xinsheng Vision, secured nearly 100 million RMB in seed funding, led by Matrix Partners China, with additional investments from Xinglian Capital, Yanyuan Venture Capital, and Shuimu Venture Capital.
- •The co-founding team includes Professor Wang Qian from Peking University's School of Psychological and Cognitive Sciences and IDG/McGovern Institute for Brain Research, Li Yuanning from ShanghaiTech University (with postdoctoral research experience at UCSF's Edward Chang Laboratory), and Professor Zhu Yixin from Peking University, an expert in embodied intelligence.
- •The startup's 'neural graphics card' system is designed for ten thousand channels and a one-piece implantation, aiming to solve the 'encoding-decoding-reconstruction' closed-loop for extremely high-bandwidth data streams in the brain.
- •Xin Sheng Shi Jie's core thesis posits that achieving stable, high-bandwidth interaction through the visual channel, the most data-intensive sensory modality, will enable a transition of BCI from a medical device paradigm to a general-purpose information interaction platform for broader human augmentation.
- •The company aims to overcome the limitations of language as a low-bandwidth, lossy compression by building a direct, high-throughput bridge between the brain's high-dimensional semantic space and large AI models and embodied AI.
📊 Competitor Analysis▸ Show
Competitor Analysis
| Company/Project | Focus | Invasiveness | Channels/Bandwidth | Key Differentiator |
|---|---|---|---|---|
| Xin Sheng Shi Jie | High-bandwidth visual & language reconstruction | Invasive | 256 channels (current 28nm chip), aiming for 10,000 channels | 'Neural graphics card' for encoding-decoding-reconstruction of high-bandwidth brain data, direct semantic connection with AI. |
| Neuralink | Vision, speech, mobility restoration | Invasive | 1,024 channels (N1 chip) | Robotic surgical system for precise implantation; fully wireless and battery-powered device. |
| Synchron | Motor function restoration | Invasive (endovascular) | 16 channels (Stentrode) | Minimally invasive endovascular implantation. |
| Paradromics | High-bandwidth neural recording | Invasive | Thousands of neural channels (Connexus®) | High-bandwidth cortical implant for streaming neural data. |
| SiClink | Visual reconstruction (bidirectional) | Invasive | Not specified | Bidirectional BCI system that both reads and writes visual information to the brain. |
| BISC (Columbia et al.) | Motor, speech, visual function restoration | Minimally invasive | Tens of thousands of electrodes, 100 Mbps wireless link | Ultra-thin, single-chip implant with high-throughput wireless data transfer. |
🛠️ Technical Deep Dive
- 'Neural Graphics Card' Concept: Xin Sheng Shi Jie's core technology is a 'neural graphics card' system designed to manage the encoding, decoding, and reconstruction of extremely high-bandwidth data streams within the brain. This system aims to become a foundational infrastructure for brain-computer integration and human-machine symbiosis.
- Chip Specifications: The company has successfully taped out 28nm BCI chips, which currently feature 256 channels. They are actively developing higher-throughput, bidirectional interactive chips. The long-term goal is to achieve a system with ten thousand channels and a one-piece implantation design.
- Visual and Language Reconstruction: The startup focuses on efficiently 'writing' visual information into the human brain and 'reading out' intentions and semantics. This addresses the functional reconstruction for patients with blindness and aphasia.
- Challenges in Visual Reconstruction: The team acknowledges the inherent difficulties in visual reconstruction, including the high-dimensional non-linear encoding of visual information, the brain's complex reliance on high-frequency saccades and internal world models, and the intricate network of lateral inhibitions and feedback connections within the primary visual cortex.
- Language Interaction: The company aims to transcend the limitations of natural language, which is viewed as a low-bandwidth, lossy compression constrained by human motor organs. The objective is to establish a direct, high-throughput semantic connection between the human brain's high-dimensional semantic space and large language models (LLMs) and embodied AI.
- Foundational Research: The team's background includes significant research, such as Professor Wang Qian's work on single-neuron recording in the visual cortex of clinical patients and studies on the mapping mechanism of color visual hallucinations induced by electrical stimulation. Additionally, research from Peking University has explored how language modulates human visual perception, using deep neural networks (like CLIP) and human brain-lesion models.
🔮 Future ImplicationsAI analysis grounded in cited sources
Xin Sheng Shi Jie will accelerate the development of general-purpose BCI platforms beyond medical applications.
By focusing on high-bandwidth visual and language reconstruction, the company aims to create a foundational BCI technology that can extend beyond medical applications to broader human-machine interaction and augmentation, positioning BCI as a future information infrastructure.
The company's technology will enable enhanced human-AI symbiosis through direct semantic interaction.
Xin Sheng Shi Jie's ultimate goal is to bypass the limitations of explicit language and actions, enabling direct, high-throughput semantic connection between the human brain's high-dimensional semantic space and large AI models and embodied AI.
Significant advancements in treating visual and language impairments will emerge from this research.
The startup's explicit focus on efficient visual information writing and semantic decoding directly addresses the critical needs of patients with blindness and aphasia, offering potential solutions where current alternatives are limited.
⏳ Timeline
2020
Professor Wang Qian's team (founder of Xin Sheng Shi Jie) published research on single-neuron recording in the visual cortex.
2026-06
Xin Sheng Shi Jie (Xinsheng Vision) founded by Peking University scientists.
2026-06
Xin Sheng Shi Jie secured nearly 100 million RMB in seed funding.
📎 Sources (10)
Factual claims are grounded in the sources below. Forward-looking analysis is AI-generated interpretation.
📰
Weekly AI Recap
Read this week's curated digest of top AI events →
👉Related Updates
AI-curated news aggregator. All content rights belong to original publishers.
Original source: 36氪 ↗
This is a summary, not the original. Read the source, or get the weekly briefing.
Weekly AI briefing
One email a week. Unsubscribe anytime.