Shenzhen Competing for 'Tsinghua Faction' Talent

marsbitPubblicato 2026-08-11Pubblicato ultima volta 2026-08-11

Introduzione

Shenzhen is actively attracting Tsinghua University-affiliated technology ventures, as highlighted during the "X-Day" Xili Lake Roadshow held in Nanshan. The event featured six startup projects from Tsinghua alumni, spanning semiconductors, AI, materials, and healthcare. The showcased companies include: Zhichen Semiconductor, developing edge AI chips; Guangsu Evolution, creating AI-powered home security systems; Qingli Technology, commercializing "self-superlubricating" technology; Shu Yu Technology, offering an AI Agent for analog chip design; Heyi Intelligent Control, providing AI-driven building management systems; and Shengshengyi, applying AI to assisted reproductive medicine. These ventures represent a trend of deep-tech innovation closely linked to academic research. The roadshow series, initiated a year ago, underscores a strategic shift in Shenzhen's investment landscape. Venture capital is moving earlier into the innovation cycle, seeking projects directly from laboratories and research papers. Tsinghua University serves as a key source for such early-stage, technology-intensive startups. Over the past two years, Tsinghua alumni projects have accounted for nearly 30% of the approximately 280 billion RMB in early-stage deep-tech funding in Shenzhen. The "X-Day" platform has facilitated significant growth. To date, its 19 roadshows have connected companies with investors thousands of times, leading to over 3.3 billion RMB in equity financing for 58 firms. Past part...

Tsinghua alumni are appearing in great numbers.

On August 6th, the "X-Day" Xili Lake Demo Event's Tsinghua Alumni Electronics & Information Special Session and the first anniversary of the South Tsinghua Public Welfare Demo were held in Nanshan, Shenzhen. Six projects took the stage, spanning chips, materials, AI, healthcare, and other fields.

One still remembers a year ago, when a batch of Tsinghua startup projects were unveiled collectively at Shenzhen's Xili Lake. Later, among them, Transcendence Intelligent (夸维智能), which emerged from the pack, recently completed a Series B financing of 10 billion RMB, with its valuation surpassing 100 billion RMB; Zero Dimension (零次方), after 18 months of establishment, completed 4 rounds of financing, with orders exceeding 100 million RMB.

Time passes, but Xili Lake always sees new faces taking the stage, and people continually move from here into the broader depths of industry.

A Group of Tsinghua Faction Appears in Shenzhen

The first to take the stage that day was Zhichen Semiconductor (智辰半导体).

This Shenzhen-headquartered company is dedicated to building a new paradigm edge AI chip platform. Founder Wang Xiaobin is a chip veteran with nearly 30 years of industry experience, having worked long-term at HiSilicon (Huawei), participating in the R&D and commercialization of multiple generations of the Kirin mobile phone chips.

Zhichen Semiconductor judges that as models continue to grow larger, more AI tasks will run locally, placing new demands on chip power consumption, storage capacity, and cost. The company is thus building the T-series edge AI chips, planning to enter scenarios like PCs, robotics, automotive, and edge servers.

The next to appear was Lightspeed Evolution (光速进化), which brought AI into the home.

Founder Li Shaozhang is a Tsinghua Computer Science alumnus with 20 years of experience in the smart hardware industry, having worked at TP-Link, 360, Meituan, and Lexar. In 2025, he started a new venture, founding Lightspeed Evolution.

The company's first product targets home security, using edge visual language models and local AI computing power to enable devices to recognize family members, strangers, pets, vehicles, and abnormal behavior, and supports functions like fall detection and safety protection.

Qingli Technology (清力技术) was founded by Academician Zheng Quanshui of the Chinese Academy of Sciences, a professor at Tsinghua University. For over 20 years, Zheng Quanshui has been researching "self-superlubricity" technology – in simple terms, achieving extremely low friction and wear when two solid surfaces slide in contact without lubricant.

Qingli Technology was officially established in 2019. Today, leveraging the "zero-wear" advantage of self-superlubricity technology, it adopts a "self-superlubricity + X" model, focusing on core technological pain points in national strategic industries and future potential tracks. It has successfully developed products like MEMS RF switches, micro-generators, and special micro-motors, and has secured small-batch orders, accelerating the provision of breakthrough solutions for fields like humanoid robots, AI computing chip testing, and 6G communication.

Later, a Tsinghua Ph.D. brought a chip design Agent. Wang Jiacheng, CEO of ShuYu Technology (枢宇科技), holds a Ph.D. in Electronic Information from Tsinghua. The company starts with analog chips – a category with a vast number of SKUs, where different application scenarios often require redesign, creating high demand for experienced analog chip engineers.

ShuYu Technology hopes to use an AI Agent to undertake part of this design work. Its core product, Orvix Pilot, uses a self-developed multi-agent framework to connect the entire circuit design process from specification input, topology generation, parameter optimization to simulation verification. It is delivered in a software-hardware integrated machine form purely for intranet use, ensuring zero data leaves the enterprise.

Heyi Intelligent Control (合一智控) grew out of the Research Center for Building Space Edge Autonomous Control at the Shenzhen Tsinghua Research Institute. The team is developing a spatial intelligence agent based on spatial perception and AI decision-making, aiming to enable buildings to perceive occupants, environment, and equipment status, and then automatically perform corresponding control.

Currently, the company's systems have entered over 200 projects, covering an area of more than one million square meters. According to the company's on-site disclosure, real-world projects have generally achieved over 20% energy savings.

The last company, ShengShengYi (生生易), brought AI into assisted reproduction. Dedicated to improving clinical decision-making, laboratory safety, and operational efficiency in fertility centers through AI software and smart medical devices, ShengShengYi has formed a product portfolio including AI embryo assessment, intelligent liquid nitrogen storage monitoring, egg retrieval vacuum pumps, and AI incubators.

Currently, the company holds two Class II National Medical Products Administration (NMPA) registrations, US FDA 510(k) clearance, and certifications for ISO 13485 and ISO 9001 quality management systems. Its products and collaborations have covered markets in China, Japan, Europe, the Middle East, and North America, continuously promoting original Chinese reproductive medicine AI technology to go global.

The six companies are at different stages of progress, yet they constitute the most authentic cross-section of Chinese technology innovation today: hard-tech startups are moving closer and closer to the source of scientific research.

From Tsinghua Garden to Xili Lake

These projects share a similar starting point: they all hail from Tsinghua.

A year ago, the first "X-Day" Tsinghua Alumni Project Special Session was held at Xili Lake, bringing together 21 Tsinghua hard-tech companies and simultaneously kicking off the first stop of the South Tsinghua Public Welfare Demo. Now, new entrepreneurial teams continue to emerge, and some projects that appeared at that time have progressed even further.

Why are Tsinghua startup projects appearing so densely here? One obvious change is that hard-tech investment is continuously extending its reach earlier. During the roundtable discussion that day, the topic most discussed by several investors was how to reach the source of technology sooner.

"This is the hottest half-year I have seen in my nearly 20-year career," said Du Yongbo, Managing Partner at Huaxing New Economy Fund, referring to the bustling scene in the venture capital circle this year.

Lin Haizhuo, Founding Partner of Zhuoyuan Asia, noted that investment is continuing to move earlier: many VC firms are entering the scientific research stage even while technological paths are still being explored. He summarized this phenomenon as the "academization" of technology investment.

Subtly, this has also changed how investment institutions find projects. "This poses more challenges for venture capital institutions, compelling us to follow the technology itself more closely and dig for projects beneath the surface," said Sui Jianfeng, General Manager of Leaguer Venture Capital. By the time a company truly enters the public eye, its financing may have already rapidly passed two or three rounds. Truly early-stage investment is moving towards professors, academic papers, and laboratories.

And Tsinghua happens to have a batch of such source projects, with electronic information being one of the densest areas. From chips and artificial intelligence to optoelectronics and smart hardware, a number of technical teams have left the laboratory and entered the stages of product definition, engineering, and commercial validation. Compared to mature companies, these projects are earlier, smaller, and also more reliant on industrial scenarios, supply chains, and the first pot of patient capital.

Over the past two years, the scale of early-stage hard-tech financing in Shenzhen has approached 280 billion RMB, with Tsinghua alumni startup projects accounting for nearly 30%; in key areas like chips, communications, and autonomous driving, the financing success rate of Tsinghua-faction projects is also significantly higher than the industry average.

Xu Ao, COO of Shuimu Nanguo Technology & Innovation (水木南国科创), introduced that the South Tsinghua Public Welfare Demo started from Nanshan a year ago, has held 17 sessions, serving 117 Tsinghua alumni startup projects. The next step will continue to link capital, industry, and policy resources, promoting more early-stage projects toward industrial implementation.

This path from universities to industry is gradually extending in Shenzhen's Nanshan District. Those entrepreneurial teams that once stood on the Xili Lake demo stage have successively grown into representative companies in their respective tracks.

To date, the "X-Day" Xili Lake Demo Club has held 19 themed demo events, cumulatively facilitating connections for about 4,500 institutions, with 58 enterprises obtaining equity financing exceeding 3.316 billion RMB. Transcendence Intelligent (跨维智能), which was on the demo stage last year, recently completed a 10 billion RMB Series B financing, with its valuation exceeding 100 billion RMB; the Tsinghua post-2000s team Zero Dimension (零次方), after 18 months of establishment, completed 4 rounds of financing, with orders exceeding 100 million RMB.

Batch after batch of projects meet at Xili Lake, and from there, they head towards more distant horizons.

This article is from WeChat Official Account "Investment Community" (ID: pedaily2012), author: Wang Lu

Domande pertinenti

QWhat is the main theme of the article titled '深圳开抢清华系'?

AThe article's main theme is the concentrated emergence of technology startups founded by Tsinghua University alumni in Shenzhen, particularly around the Xili Lake area. It highlights how these 'Tsinghua system' projects are attracting investment and accelerating their growth within Shenzhen's innovation ecosystem.

QWhat event is described at the beginning of the article, and what was its focus?

AThe event described is the 'X-Day' Xili Lake Roadshow Society's Tsinghua Alumni Electronic Information Special Session, held in Nanshan, Shenzhen on August 6th. It focused on showcasing six Tsinghua-affiliated startup projects spanning fields like chips, materials, AI, and healthcare.

QName two Tsinghua-affiliated startups mentioned in the article that have shown significant growth after participating in the Xili Lake roadshows.

ATwo such startups are Kuawei Intelligent (跨维智能), which recently completed a 10 billion RMB Series B financing with a valuation over 100 billion RMB, and Lingcifang (零次方), which secured 4 rounds of financing in 18 months and achieved over 100 million RMB in orders.

QAccording to the investors quoted in the article, what trend is occurring in venture capital investment for hard tech?

AInvestors note a trend of investment moving earlier into the research and development cycle. This 'scientification' of tech investment means VCs are engaging with projects at the laboratory or even theoretical research stage, seeking to identify and fund promising technologies at their source.

QWhat role does the 'Nan Guo Tsinghua Public Welfare Roadshow' play, according to the article?

AThe 'Nan Guo Tsinghua Public Welfare Roadshow' serves as a platform to connect Tsinghua alumni startups with capital, industry resources, and policy support. Having held 17 sessions and served 117 projects in one year, it aims to facilitate the transition of early-stage projects from the lab to industrial application and commercialization.

Letture associate

What's New in Ethereum's Roadmap: Privacy, Quantum Security, Native Rollup?

On August 10, 2024, Ethereum co-founder Vitalik Buterin revealed an updated technical roadmap, highlighting significant changes in priorities and new focus areas. Key new additions include: - **Strong Privacy Protection**: Elevated to a top-tier protocol concern. Proposals like EIP-8250 (keyed nonces for concurrency), EIP-8272 (recent roots for verification), and EIP-8182 (shared protocol-level privacy pools) aim to enable private transfers of ETH and ERC-20s, moving toward default private accounts. - **Quantum Resilience**: Priority increased. The plan addresses risks to ECDSA/BLS signatures, KZG commitments, and ZK systems, exploring solutions like leanSPHINCS signatures, STARK proofs (considered quantum-resistant), and streamlined verification processes. A full L1 upgrade is tentatively targeted for ~2029. - **Native Rollups**: A novel concept where rollups could reuse Ethereum's core validation infrastructure via a new EXECUTE precompile (EIP-8079 draft). This aims to reduce custom code and enhance security for certain L2s, while specialized chains will continue to exist. - **Protocol Simplification & Formal Verification**: Leveraging AI tools to make full-protocol formal verification feasible, aiming for more rigorous and machine-verifiable specifications. Other notable shifts: - State storage is evolving from Verkle trees to a Partitioned Binary Tree (PBT, EIP-8347 draft). - Some 2023 items like VDFs and certain EVM improvements were deprioritized. - Long-term discussions include blob/gas futures and non-EVM instruction sets (e.g., RISC-V). The "Strawmap" outlines potential upgrades through ~2029 but is a directional guide, not a firm timeline. Key milestones to watch include the Glamsterdam (Q4 2026) and Hegotá (2027) forks. Many proposals remain in draft or research phases.

marsbit7 min fa

What's New in Ethereum's Roadmap: Privacy, Quantum Security, Native Rollup?

marsbit7 min fa

Hyperliquid Trading Volume Soars, So Why Are Profits Falling?

Hyperliquid, a leading decentralized perpetuals trading platform, has seen its open interest surge to a record high above $11 billion, capturing roughly 9% of the global market share. Trading volume remains robust, nearing $178 billion over 30 days, driven largely by the explosive growth of third-party markets offering tokenized real-world assets (RWAs) like stocks and commodities. Despite this growth, the platform's protocol revenue has declined for four consecutive quarters, falling 43% from its Q3 2025 peak of $357 million to approximately $202 million in Q2 2026. This divergence is primarily attributed to the HIP-3 governance proposal, which allows external developers to launch their own markets and keep up to half of the generated fees. These third-party markets now account for nearly 50% of total volume. Consequently, the share of revenue redistributed to developers, market makers, and the treasury has tripled from 6% to 18% in a year. This directly reduces the funds allocated to the platform's buyback-and-burn mechanism for its native HYPE token, weakening a key price support. HYPE's price has fallen 28% from its all-time high. The platform's growth is also heavily concentrated, with a single entity, Trade.xyz, responsible for over 90% of HIP-3 open interest, introducing systemic risk. Additionally, the ecosystem lacks diversity beyond HYPE, faces ongoing token unlocks adding sell pressure, and is encountering increased regulatory scrutiny and new competition from platforms like Robinhood Chain. While still a major revenue generator in crypto, Hyperliquid's model of sharing fees to fuel expansion is currently compressing its own earnings and token economics.

marsbit7 min fa

Hyperliquid Trading Volume Soars, So Why Are Profits Falling?

marsbit7 min fa

Citi Research Report Analysis: U.S. Proposed Ban on Chinese Optical Modules Has No Substantial Progress, Short-term Enforcement Faces Supply Constraints

Citi Report Analysis: U.S. Proposed Ban on Chinese Optical Modules Lacks Substantive Progress, Faces Supply Bottlenecks in Short Term. Reuters reported on August 4th that the U.S. government and FCC are considering a ban on Chinese optical modules. Citi's August 9th report clarifies that optical modules are not listed on any effective FCC ban. The FCC's Order 26-50 established two restricted list mechanisms (based on manufacturer and production location), but optical modules were only mentioned once, as an example in a disclosure requirement, not as a restricted product. The reported ban remains at a proposal stage. Citi estimates Chinese suppliers provide 60-70% of high-speed optical modules for U.S. hyperscalers. Non-Chinese suppliers cannot fill this gap in the short term, making the immediate implementation of a genuine ban unlikely. Future regulatory paths could be manufacturer-based (least likely), location-based covering all offshore production (strictest), or location-based covering only China (more feasible but with unresolved definitions). A ban would pressure U.S. AI infrastructure, conflicting with stated policy goals. Citi sees low near-term implementation probability, with the issue potentially becoming a negotiation chip in bilateral talks. U.S. domestic capacity build-out is a key long-term variable. Among Chinese companies, XSENS and Dongshan Precision have the highest U.S. exposure, while Tianfu Communication, as a passive component supplier, is relatively insulated. Citi maintains Buy ratings on all three with respective price targets. The conclusion is that Chinese modules are currently irreplaceable in the U.S. AI supply chain, creating a longer timeline for potential restrictions than the market may expect.

marsbit1 h fa

Citi Research Report Analysis: U.S. Proposed Ban on Chinese Optical Modules Has No Substantial Progress, Short-term Enforcement Faces Supply Constraints

marsbit1 h fa

Trading

Spot
活动图片