• Phosio

  Participated · Seed · Jun 2026

  Phosio develops high refractive index metal oxides and a proprietary thin-film thin-film platform designed to integrate display functionality directly into everyday eyewear. The technology aims to reduce cost, bulk, and power demands of current AR display systems to enable all-day wearable glasses at consumer-friendly price points. The company was co-founded by CEO Omid Sadeghi, Cory Perkins, and materials scientist Douglas Keszler and brings together expertise in advanced materials science, optics, and semiconductor manufacturing. Phosio highlights team experience that includes former Meta personnel and members of the early scale-up team from Inpria. Financially, Phosio recently raised a $4 million seed round led by MESH with several strategic investors to accelerate commercialization. The company is engaging partners across the semiconductor and consumer electronics ecosystems as it moves toward commercial production.

• Neurophos

  Participated · Series A · Jan 2026

  Neurophos is an Austin-based semiconductor startup building optical processing units (OPUs) that integrate more than one million micron-scale optical processing elements onto a single chip. Using proprietary metamaterial optical modulators—miniaturized 10,000-fold versus prior photonic components—the company targets up to a 100× gain in both speed and energy efficiency over leading silicon GPUs. Its technology is positioned as a drop-in replacement for current data-center accelerators, addressing power, scalability and cost constraints facing large-scale AI inference. Total capital raised now stands at $118 million, providing resources to commercialize the first integrated photonic compute system, complete with datacenter-ready OPU modules, software stack and developer hardware. The Series A proceeds will support expansion of the Austin headquarters and the opening of a San Francisco engineering office to serve early customers. Neurophos’ founding team includes veterans from NVIDIA, Apple, Intel and other major chipmakers, bolstering its credibility in bringing photonic computing to market.

• Quadric

  Participated · Series C · Jan 2026

  Quadric develops the Chimera GPNPU, a fully programmable general-purpose AI inference processor architecture optimized for on-device AI/ML inference and classic DSP and control algorithms. The Chimera processors scale from 1 to over 3,200 TOPS in multi-chiplet configurations and support both convolutional and transformer-based models, including on-device LLM inference and emerging VLA world models. Quadric's software toolchain converts AI models into C++ and enables SoC teams to program in Python or C++, allowing models to be updated on shipped silicon. The company offers automotive-grade safety-enhanced versions of its architecture and emphasizes a software-updatable approach so the same silicon can run models published after the chip was designed. Quadric reported that product revenues more than tripled in the prior year and that it reached profitability. The company is headquartered in Burlingame, California.

• Niobium Microsystems

  Participated · Equity · Dec 2025

  Niobium Microsystems is a Dayton-based company developing specialized chips that bring fully homomorphic encryption (FHE) into practical commercial use. Its custom silicon architecture allows data to remain encrypted even while it is being processed, offering mathematically guaranteed privacy for customers handling sensitive information. Led by CEO Kevin Yoder, the firm is currently building a production-ready application-specific integrated circuit (ASIC) and enhancing supporting infrastructure for pilot integrations. With offices in Portland and San Francisco, Niobium plans to broaden ecosystem enablement so partners can more readily adopt FHE-based solutions. The company recently secured more than $23 million in fresh capital to finalize its production silicon design and bolster customer-facing capabilities. Although financial performance metrics such as revenue or user count were not disclosed, the size of the latest raise underscores investor confidence in the company’s technology roadmap.

• Lemurian Labs

  Participated · Series A · Dec 2025

  Headquartered in San Francisco and led by CEO Jay Dawani, Lemurian Labs has built a platform that combines compiler technology with runtime orchestration to let developers write AI code once and deploy it on any hardware. Its hardware-agnostic approach promises faster deployment cycles, greater flexibility and lower infrastructure costs at scale. The company is positioning the platform for organizations seeking sustainable compute and open AI innovation. Fresh capital will be directed toward expanding the engineering team, accelerating product development and deepening partnerships across the AI ecosystem. Although specific revenue or user figures were not disclosed, the company previously completed a seed round in 2022. Lemurian Labs believes its technology can reduce the friction of moving AI workloads between edge devices, cloud instances and on-prem servers, potentially widening its addressable market.

No current team members are available.