CONNECT WITH US
Monday 29 December 2025
MetAI to debut MetGen at CES 2026, targeting the US market
In recent years, digital twin technology has quickly gained traction across industries. With the rapid maturation of AI and big data analytics, enterprises can now build highly accurate, real-time synchronized virtual models, enabling predictive maintenance, process optimization, and other benefits that were once difficult to realize. MetAI, invited by Taiwan Tech Arena (TTA) to join CES 2026, will showcase its proprietary AI - and 3D-driven MetGen digital twin platform, formally signaling its intent to capture opportunities in the US smart warehousing sector.MetAI co-founder and CEO Daniel  Yu noted that the US is home to some of the world's largest warehousing market, driven by digitalization efforts from major retailers such as Amazon, Walmart, and Costco. As such, the US has become MetAI's primary target for overseas expansion. "We are honored to receive TTA's invitation to exhibit at CES 2026," Yu said, expressing appreciation for TTA's comprehensive support - from pre-show preparation and customer matchmaking to presentation guidance - which has strengthened the company's confidence in entering the US market.AI fused with 3D technology : Digital twins generated in minutesFounded in 2023, MetAI focuses on developing MetGen, the first AI-native, domain-specific generative platform designed for real-to-sim and sim-to-real integration. By leveraging its proprietary AI and 3D synthesis architecture, MetGen automatically converts traditional CAD and 2D design files into SimReady (simulation-ready) 3D digital twin environments. The platform also incorporates AI, 3D simulation, physics modeling, and automation control logic, enabling users in warehousing, advanced manufacturing, semiconductors, and automation equipment to design, validate, and simulate workflows in a virtual environment before real-world deployment.MetAI further supports the creation of large-scale, high-fidelity synthetic datasets within the digital twin environment, giving AI models access to virtually unlimited training material. This capability accelerates production-line design and product development, while enabling enterprises to derive optimized strategies in simulation before applying them in real operations.By integrating NVIDIA Omniverse with MetAI's proprietary generative models, complex digital twin environments that once required months or even years to build can now be automatically generated in minutes, effectively resolving one of the industry's most persistent adoption barriers: excessively long deployment timelines. The company currently focuses on three key application areas - smart warehousing, semiconductors, and data centers - and completed a US$4 million seed round in 2025, backed by leading investors including Kenmec Mechanical and NVIDIA, making it one of the few Taiwan startups to receive a direct strategic investment from NVIDIA.According to Yu, traditional digital twin development relies heavily on manual data processing, resulting in implementation cycles that may stretch across several months or even longer than a year. Even with newer solutions emerging, model inaccuracies often lead to costly revisions. "MetAI combines AI with 3D generation to directly interpret 2D blueprints such as CAD or BIM files and automatically construct complete 3D twin environments," Yu explained.This approach reduces the typical modeling workload - from hundreds or thousands of hours - to just a few minutes, while delivering high consistency and zero structural error. It enables companies to rapidly plan production lines, build virtual sandboxes for robot training or logistics flow optimization, and significantly boost deployment efficiency.Establishing a US headquarters to capture smart warehousing and advanced manufacturing opportunitiesDespite being a relatively young company, MetAI has quickly secured partnerships with several high-profile enterprises, thanks to the uniqueness and technical depth of its solutions. In addition to working with TSMC, MetAI is collaborating with Kenmec to fully reconstruct the Chief Global Logistics's smart logistics center  into a virtual environment, creating a highly realistic digital twin featuring physics-accurate modeling, executable control logic, and AI-driven optimization capabilities.According to Yu, the rise of AI has prompted Taiwanese startups to shift their mindset - no longer focusing solely on the domestic market but aiming directly at the world's largest and most innovation-driven arenas. "The US has exceptionally strong demand for smart warehousing solutions. We are actively preparing to establish our US team and headquarters, which will include a local team, customer support, and a technical service center. This will place us closer to key customers and significantly enhance our ability to execute in the North American market," Yu said.Yu added that MetAI will continue to strengthen its real-to-sim and sim-to-real capabilities, while expanding integration with a broader ecosystem of automation hardware. The company aims to build an end-to-end twin platform that spans design, simulation, and deployment. Ultimately, MetAI seeks to enable enterprises to advance smart warehousing and smart factory transformations at minimal cost and with maximum speed.MetAI partners with Kenmec to build a high-fidelity virtual smart logistics center. Credit: MetAI
Monday 29 December 2025
BOS Semiconductors Presents AI Box for Mobility at CES 2026
SEOUL, South Korea—December 24, 2025—BOS Semiconductors, a fabless semiconductor company specializing in automotive and physical AI semiconductors, today announced that it will participate in CES 2026, the world's largest technology exhibition, taking place in Las Vegas, USA from January 6–9, 2026. At the event, BOS will unveil an AI Box demo designed for next-generation mobility.At CES 2026, BOS will highlight its strategy aligned with major industry trends, including the advancement of autonomous driving, the shift toward Software Defined Vehicles (SDVs), and the expansion of Physical AI. Through the demonstrations, BOS will present how to anticipate emerging needs in mobility AI environments - such as: Support for a wide range of AI models based on CNN and Transformer architectures, Flexible scalability and integration with existing automotive electronic systems, Real-time perception and decision-making enabled by physical AI.To support these capabilities, BOS will showcase AI model demos powered by its AI Box integrating the high-performance AI accelerator Eagle-N. The AI Box is an external AI computing module that allows automakers (OEMs) to add an "AI brain" to vehicles without replacing existing in-vehicle infotainment (IVI) systems, enabling fast AI expansion with minimal platform change.With the AI Box, OEMs can minimize changes to their existing systems while extending high-performance AI functionality independently. This approach accelerates adoption of advanced AI capabilities not only in new vehicles but also in facelift models - updated versions of existing vehicles with enhanced design and features - reducing development cost and time while improving long-term product competitiveness.The AI Box is based on an on-device AI architecture, delivering several advantages. Sensitive data such as voice and video can be processed directly inside the vehicle rather than being transmitted to the cloud, strengthening privacy protection and data security. It also ensures stable AI operation regardless of network connectivity, improving overall reliability. Over the long term, the AI Box can reduce total cost of ownership (TCO) by lowering cloud traffic, inference, and storage costs. In addition, it offers strong scalability without requiring server expansion as the number of users increases.BOS' AI Box is designed to integrate flexibly with existing vehicle systems through multiple interfaces, enabling an efficient AI vehicle architecture based on clear role separation: AI-intensive functions are handled by the AI Box, while existing systems continue performing their original roles. At the booth (Venetian Expo, Hall A, Booth #50017), BOS will demonstrate on-device AI models that combine Vision-Language Models (VLMs) and Large Language Models (LLMs) based on the AI Box architecture. The demo will also be showcased at the Tenstorrent Demo Room as well."Through our AI Box demo with Eagle-N at CES 2026, we will present a practical approach for effectively scaling AI capabilities even on existing mobility system," said Jason (Jeongseok) Chae, Vice President and Head of Strategic Marketing & Sales at BOS. "We aim to evolve beyond automotive semiconductors and become a core semiconductor company leading the era of Physical AI."
Monday 29 December 2025
Global IT and Innovation: Kyrgyz Republic at CES 2026
The High Technology Park of the Kyrgyz Republic (HTP) will participate in CES 2026, presenting the country's growing strength in artificial intelligence, digital health, blockchain technologies, and large-scale IT exports for global markets. At Booth 50265, the Kyrgyz delegation will showcase technology companies delivering AI products, telemedicine platforms, blockchain infrastructure, and software development services for clients worldwide.The Kyrgyz Republic has become one of the fastest-growing technology exporters in its region. In 2024, Kyrgyz IT specialists provided services to clients in more than 63 countries, generating $130 million in IT exports. The United States ranked first, accounting for around 40 percent of total exports, with an export volume exceeding $50 million, including cooperation with companies in Silicon Valley. Over the past five years, the country's IT exports have grown 45-fold, marking a structural shift toward high-value digital services and proprietary technology development. HTP now brings together a network of 500 software development companies and 3,000 developers, ready to meet the full range of client needs.Among the technologies presented at CES 2026 is KaniTTS, an open-source text-to-speech model developed by the AI startup NineNineSix. KaniTTS redefines real-time, human-like voice generation, matching the performance of leading commercial voice models while remaining fully open and free under the Apache 2.0 license. Since its release, KaniTTS has been downloaded more than 15,000 times on Hugging Face, drawing international attention for its speed, expressivity, and accessibility. The model is capable of generating 15 seconds of natural speech in just one second on a consumer-grade NVIDIA RTX 5080 GPU, enabling true real-time performance without cloud-scale infrastructure. The base model currently supports eight languages: English, German, Korean, Arabic, Chinese, Spanish, Kyrgyz, and Japanese.In parallel, NineNineSix together with HTP has released Kyrgyz Whisper, an open-source automatic speech recognition model fine-tuned on Whisper to deliver native speech recognition for Kyrgyz, Russian, and English. Performance highlights include a dramatic improvement in Kyrgyz word error rate from approximately 100 percent to 0.2 WER, while maintaining stable performance in English and Russian. Two model variants are available, a small version with 0.2 billion parameters and a medium version with 0.8 billion parameters, both supporting LoRA fine-tuning for domain adaptation and further improvements.The Kyrgyz delegation also includes Blockchain Data, a blockchain technology company developing digital infrastructure and crypto-related solutions. The Kyrgyz Republic has recently completed a major milestone in the digital finance sector with the launch of its national stablecoin, KGST, pegged 1:1 to the Kyrgyz som. KGST complies with international KYC/AML standards, features multi-level authentication, and has completed technical testing for deployment on both regional and global crypto platforms. According to the Chainalysis Global Crypto Adoption Index, the Kyrgyz Republic ranks 19th globally in cryptocurrency usage, reflecting an active, regulated, and rapidly developing crypto ecosystem.Digital health innovation will also be represented through Hi Doctor, an AI-powered telemedicine and clinical decision-support platform founded by diagnostic physician Nursultan Masalbekov. The product emerged from real clinical practice, where the lack of digital tools forced doctors to spend excessive time on documentation instead of patient care. Hi Doctor combines AI-driven analysis of clinical data, a voice assistant for medical documentation, and telemedicine tools, including NORAi, which analyzes patients' vital signs during video consultations. The platform is actively used by doctors and medical institutions across Kyrgyzstan and Kazakhstan. In multiple cases, AI-supported recommendations contributed to the early detection of HIV, cardiomyopathies, oncological and genetic conditions, later confirmed through laboratory testing, demonstrating the platform's real clinical impact.The Kyrgyz Republic invites attendees to meet its delegation at CES 2026 at the High Technology Park of the Kyrgyz Republic, Booth 50265, from January 6 to 9, 2026, in Las Vegas, Nevada.