11 Startups Won the Fourth Batch of IC Taiwan Grand Challenge - Awards Focused on Innovative IC Solutions and Applications

According to Deloitte, the global semiconductor industry is expected to reach US$975 billion in annual sales in 2026, with 22% growth rate in 2025 and is projected to accelerate to 26% in 2026. Another report by McKinsey projected that the semiconductor industry could reach USD 1.6 trillion as early as 2030, primarily fueled by the AI boom which includes new computing models, domain-specific architectures, advanced chip sets, and other next-generation technologies. Leveraging this projected massive growth, the National Science and Technology Council (NSTC) continues to promote the IC Taiwan Grand Challenge (ICTGC). Now in its fourth batch, the competition seeks to nurture talented global tech startups in the highly competitive, rapidly growing, and vital semiconductor industry.

Announced on May 6th, 2026, the winners of ICTGC batch 4 represented 7 countries and regions, including: Taiwan, US, Japan, UK, Germany, France, and Israel. The winners leveraged their innovative solutions in AI core technologies and chips, sustainability, smart manufacturing, and smart mobility. The winning startups are:

ThermoVerse Inc. - Building the Thermal Muscle of Smart Cities
ThermoVerse is a US-based startup that developed LATCHES, a programmable, zero-footprint solid-state battery for HVAC peak load management. Their zero-footprint, PCM-based rechargeable battery reduces HVAC energy use by 20–60%, freeing up more power without sacrificing human comfort and allowing users to utilize the power for other energy intensive functions including EV charging or to power up AI infrastructure.

Oculi - See Everything, Store Nothing, Save Energy
The US-based Oculi redefines Vision AI at the edge by maximizing efficiency, preserving privacy, and enabling real-time intelligence. Their IntelliPixel® turns each pixel into a smart, programmable sensing unit that processes only essential data, dramatically reducing power consumption, bandwidth requirements, and system costs while delivering ultra‑low‑latency performance for always‑on AI, enabling visually smarter devices, machines, and environments.


ROCKCORE TECHNOLOGY CO., LTD. - Delivering Ultra-Precise AMD+FPGA Heterogeneous Computing
Based in Taiwan, ROCKCORE TECHNOLOGY develops AMD+FPGA heterogeneous systems (AI-FPC) for defense and high-end equipment. Focusing first on IC substrates, ROCKCORE TECHNOLOGY specializes in AI-FPC hardware design, FPGA software, hardware, and firmware system integration, as well as FPGA IP and platform development which are mainly applied in high-end defense, IC substrate, and semiconductor equipment.


TopoLogic Inc. - High-Capacity Sub-Ns-Speed Memory For Gb Capacity Last Level Cache Memory
TopoLogic is a Japan-based developer of TL-RAM, which uses non-collinear topological antiferromagnetic memory to create a magnetic memory capable of sub nanosecond speeds and GB capacity. Their technology allows 10x capacity increase of last level cache in high performance logic chip processors, greatly enhancing the computing capability of GPUs, CPUs and TPUs by enabling larger program processing without the latency of accessing outside-of-chip memory.


Caesarea Labs - Semiconductor EDA for Post-Silicon Validation Labs
Caesarea Labs is an Israeli startup that will shift circuit edit from a fragile, slow workflow to a reliable, repeatable, fast industry standard. They are building the first end-to-end automation of the Circuit Edit workflow including automated translation of solution hypotheses into layout-level CE designs, elimination of manual layout navigation and annotation, advanced 2D/3D visualization of edits directly within the chip database, and AI-assisted CE design generation, scoring, and alternative exploration.


Infiniflux Ltd. - Keeping AI Cool
Infiniflux is a UK-based deeptech startup redefining data centre efficiency. Their patent-pending 2-phase Direct-to-Die liquid cooling system delivers industry-leading performance, keeping GPUs 15-20ºC cooler, even at >1000W per chip, resulting in up to 15% more compute capacity, much lower energy & maintenance costs, and over 2x longer chip lifespan compared with the best 1-phase cooling systems.


Linque - Transforming AI Infrastructure with Photonics
Linque is a German-based startup aiming to transform AI infrastructure with photonic network switches for data centers. Linque RISE is an all-optical circuit switching technology built on large-scale photonic integrated circuits that routes data entirely in the optical domain, eliminating power-hungry optical-electrical-optical conversions. Linque’s technology enables ultra-fast, energy-efficient all-optical switching through large-scale photonic integrated circuits.


Aniah - Delivering Transistor-Level Verification for Analog SoCs
Based in France, Aniah is an EDA company addressing transistor-level verification for advanced analog and mixed-signal SoCs. Their core technology combines a fast, vector-less transistor-level verification engine with an LLM-based design intelligence layer. Aniah’s OneCheck performs exhaustive, vector-less electrical rule checking across all power states and voltage domains, catching critical corner cases missed by simulation.


STR8 Industries - Developing Edge-Native Autonomy Software and Modular Sensing Architectures
STR8 Industries is a US-Taiwan venture-backed maritime autonomy company building autonomous navigation software and integrated onboard systems. STR8 Industries' core technology is an edge-native autonomy stack that ingests data from a distributed multi-node sensor suite to create a unified real-time world model. The system continuously tracks and prioritizes contacts, computes collision risk (CPA/TCPA), and generates COLREGs-compliant avoidance maneuvers with full audit logging.


Leafy Lab Inc. - Bridging Silicon Physics and AI to Turn Chip Design Guesswork Into Day Long Precision
Leafy Lab is a US-based startup focusing on applied AI platforms transforming analog and mixed-signal chip design. By bridging silicon physics with physical AI, their explainable models replace month-long trial-and-error with day-long precision workflows. Leafy Lab’s generative copilot predicts device behaviors with ~2% error and automates complex layouts, empowering chip designers to optimize performance, aim to cut time-to-market by 50%, and achieve first-pass silicon success.

Vellex Computing - High-Speed, Low-Power Edge Computers for Personalized, Learning AI


A US-based startup, Vellex Computing spearheads an AI paradigm shift with high-speed, ultra-low-power edge computers to democratize high-performance AI. Vellex Computing’s architecture achieves GPU-level processing with 100x less energy, enabling real-time AI training for wake-word or anomaly detection locally on battery-powered wearables and sensors without cloud reliance. By bridging compute demand and efficiency, Vellex Computing unlocks the next generation of always-on intelligent apps.

Meet the winners of ICTGC batch 4 at InnoVEX 2026!

Winners of the ICTGC batch 4 will showcase their innovations in InnoVEX 2026 “IC Taiwan Grand Challenge Award Ceremony & Unveil,” and in ICTGC Pavilion. Register for InnoVEX 2026 and ensure you have a valid visitor badge first for InnoVEX 2026 entry and connect with the winners. Details are as follows:


IC Taiwan Grand Challenge Award Ceremony & Unveil

Event date: June 2, 15:00-17:30 (UTC+8)

Event venue: Taipei Nangang Exhibition Center (TaiNEX) Halls 2 4F, Center Stage at InnoVEX

Sign up here: https://innovex.computex.biz/SHOW/forum.aspx?ID=102


*Please ensure you have a valid visitor badge first for InnoVEX 2026 entry. (Age 18+ only).

Apply for your visitor badge here: https://www.computexonline.com.tw/