Key Takeaways
- • CAS Star's decade-long photonics thesis validated: Mi Lei, founder of the Chinese venture capital firm born from the state-run Chinese Academy of Sciences, has deployed capital across more than 200 portfolio companies spanning photonics sectors (sensing, communications, computing, storage, display) and is now capturing significant AI-driven returns as the industry recognizes optical interconnects as essential infrastructure.
- • Nvidia accelerating photonics investment at scale: Nvidia has committed at least $4–6.5 billion since March 2026 to companies developing photonics technology, including major investments in Coherent (market cap $51 billion) and Lumentum, signaling that optical interconnects are now critical to solving AI data-center bottlenecks.
- • Copper interconnects hitting physical limits in AI clusters: As AI model training scales to tens of thousands of specialized chips requiring constant data sharing, traditional copper cables create signal loss, excessive heat, and power consumption constraints that optical links—using light instead of electricity—now solve at scale.
- • Photonics market valued at $154 billion: The AI photonics sector represents a $154 billion addressable market, with Nvidia-backed firms and lesser-known optical suppliers positioned to capture outsized growth as data centers transition to light-based interconnect architecture.
- • Early positioning advantage for contrarian investors: CAS Star's sustained, decade-long commitment to photonics before mainstream institutional capital arrived demonstrates the financial upside of conviction-based thematic investing in deep technology before market recognition.
Summary
CAS Star, a venture capital firm founded by Mi Lei and born from China's state-run Academy of Sciences, is reaping substantial returns from a contrarian, decade-long thesis on photonics technology that has only recently entered mainstream institutional consciousness. Mi Lei, who holds a doctorate in optics from the Xi'an Institute of Optics and Precision Mechanics under the Chinese Academy of Sciences, began investing in photonics from CAS Star's inception. The firm now manages a portfolio exceeding 600 companies, with more than 200 focused on photonics across sensing, communications, computing, storage, and display applications. This early positioning reflects a strategic conviction that scientific research in optics would eventually drive industrial-scale commercialization—a thesis that AI's explosive infrastructure demands have now validated.
The catalyst for photonics' sudden global revaluation stems from fundamental physics constraints in artificial intelligence infrastructure. As AI model training scales to unprecedented sizes, data centers must now coordinate tens of thousands of specialized computing chips in constant communication. Traditional copper interconnects, the backbone of legacy data-center architecture, are hitting hard physical limits: transmitting large data volumes through electrical signals causes signal degradation, generates excessive heat, and drives power consumption to unsustainable levels. Optical links—which transmit data using light instead of electricity—eliminate these bottlenecks and have become central to modern data-center design. This shift from electrical to optical interconnects represents a structural, long-term requirement rather than a cyclical trend.
Nvidia's recent capital commitments underscore the magnitude of this transition. Since March 2026, Nvidia has invested at least $4–6.5 billion in photonics supply chain companies, including major positions in Coherent (currently valued at $51 billion market capitalization) and Lumentum. These investments target optical interconnect manufacturers critical to scaling AI clusters. The AI photonics market itself is now valued at approximately $154 billion, with Nvidia-backed firms and emerging optical suppliers positioned to capture substantial growth as data centers globally rebuild their interconnect architectures around photonic standards.
Mi Lei's framing of CAS Star's mission—advancing "scientific research towards commercialisation"—captures the essence of the firm's decade-long positioning. The venture capital office in Shanghai, notably named "Wallfacer" after Liu Cixin's sci-fi strategists, reflects a conviction-driven approach to identifying technology inflection points before broader market recognition. For institutional investors, CAS Star's trajectory demonstrates the financial significance of thematic discipline: sustained capital deployment in deep-technology areas, grounded in scientific understanding and patient for commercialization, can yield outsized returns once market structure shifts to validate the original thesis.
The catalyst for photonics' sudden global revaluation stems from fundamental physics constraints in artificial intelligence infrastructure. As AI model training scales to unprecedented sizes, data centers must now coordinate tens of thousands of specialized computing chips in constant communication. Traditional copper interconnects, the backbone of legacy data-center architecture, are hitting hard physical limits: transmitting large data volumes through electrical signals causes signal degradation, generates excessive heat, and drives power consumption to unsustainable levels. Optical links—which transmit data using light instead of electricity—eliminate these bottlenecks and have become central to modern data-center design. This shift from electrical to optical interconnects represents a structural, long-term requirement rather than a cyclical trend.
Nvidia's recent capital commitments underscore the magnitude of this transition. Since March 2026, Nvidia has invested at least $4–6.5 billion in photonics supply chain companies, including major positions in Coherent (currently valued at $51 billion market capitalization) and Lumentum. These investments target optical interconnect manufacturers critical to scaling AI clusters. The AI photonics market itself is now valued at approximately $154 billion, with Nvidia-backed firms and emerging optical suppliers positioned to capture substantial growth as data centers globally rebuild their interconnect architectures around photonic standards.
Mi Lei's framing of CAS Star's mission—advancing "scientific research towards commercialisation"—captures the essence of the firm's decade-long positioning. The venture capital office in Shanghai, notably named "Wallfacer" after Liu Cixin's sci-fi strategists, reflects a conviction-driven approach to identifying technology inflection points before broader market recognition. For institutional investors, CAS Star's trajectory demonstrates the financial significance of thematic discipline: sustained capital deployment in deep-technology areas, grounded in scientific understanding and patient for commercialization, can yield outsized returns once market structure shifts to validate the original thesis.
Go deeper than the headline
Research every firm in this story, track the ones that matter and get the daily briefing — free for 7 days.