In the past 10 years, NEXCOM has transitioned from a hardware-centric company to a provider of comprehensive AIoT solutions across edge computing, robotics, and cyber security. Where does the company stand today in terms of products, manufacturing locations, and main customers?
Dr. Ooi: We not only sell hardware but also help implement complete solutions in manufacturing, industrial automation, healthcare, and legal and education sectors.
We handle multiple layers—from operational technology to enterprise-level visualisation systems, like big screens showing factory-wide operations. In factory automation, for example, we provide devices and help integrate them with software. All our hardware is produced in our two factories in Taiwan, and we sell to customers in Europe, the West, and Taiwan.
Mr. Lin: In Greater China, we also provide smart manufacturing consulting. We believe factories should have no boundaries, and we help transition traditional factories into digital ones. We do not just offer bundled hardware and software—we go on-site, interview clients, and tailor proposals. If they proceed, we offer full system integration services.
NEXCOM's latest AI tools for industrial applications are designed to simplify AI use on the factory floor. In a world dominated by cloud-based systems, what advantages do edge-computing tools like the LLM-powered assistant EdgeGPT provide?
Dr. Ooi: Locally placed systems offer advantages. First, data stays local, enhancing privacy and security—crucial in manufacturing. Second, computing close to the data source enables faster processing, essential in smart mobility. An autonomous vehicle cannot wait on cloud instructions to stop or turn. Even with 5G, latency remains an issue. Lastly, edge computing is cost-effective when only a small AI model is needed. You do not need massive cloud infrastructure to make intelligent decisions.
Mr. Lin: I prefer the term "hybrid" computing—using cloud or edge depending on the application. For fast-response or privacy-sensitive scenarios, you use on-premise edge systems like EdgeGPT. But for training AI models and storing historical data, we use cloud platforms like AWS or Azure. It is all about balance.
One of NEXCOM’s goals is to enable a collaborative AI ecosystem. But we are seeing major players like Microsoft internalise money-making AI services in proprietary data centres. Are you seeing this shift too? What do you make of this trend?
Dr. Ooi: Big companies want to protect their core assets—things developed through years of R&D. But others want access, and governments have a role. In Taiwan, for instance, the government supports small companies and start-ups to access data and computing resources. I believe sharing will continue because open-source collaboration has driven AI progress so far.
Still, companies need to protect their bread and butter. It is not black and white. I think we will see both: more sharing where possible, more protection where needed.
Let’s talk about robotics. The NexMOV2 is an autonomous mobile robot with AI-powered 3D vision that eliminates the need for expensive LiDAR. Where is this technology deployed?
Dr. Ooi: We use it in our own factory. Affordable AMRs typically use 2D LiDAR, unlike autonomous vehicles that use advanced 3D LiDAR. With 2D, you can only detect and align to obstacles, not understand them. Our second-generation NexMOV2 is more intelligent, capable of identifying objects and navigating dynamically—important in factories with human workers.
It is also used in unmanned hotels. A robotic arm stores guest luggage, and when the guest is in their room, they use a local voice assistant to request delivery. The AMR collects the correct container, calls the elevator, rides to the correct floor, stops at the door, and notifies the guest. The guest enters a password to receive the luggage, and the robot returns. It integrates arms, elevators, and voice systems.
That sounds like a highly complex set of tasks. I imagine generative AI comes into play when interacting with the Alexa-based system?
Dr. Ooi: The Alexa-type system is part of the external infrastructure, not built into our AMR. We are working on a related project, but that is confidential for now.
How advanced are the chip-requirements for this system?
Dr. Ooi: Our AMRs are battery-run, so power consumption versus performance is key. We looked at how much computation we could get per watt of CPU power. A GPU could handle AI workloads but would drain the battery quickly. So, we do not necessarily use the most high-end CPUs. It is about implementing the software efficiently. We offer both NVIDIA and Intel platforms so clients can choose what fits their needs.
What other EdgeAI advancements set NEXCOM apart in 2025?
Dr. Ooi: Functional robotic safety is a key differentiator. We worked with Intel on a reference design board for a functional safety solution based on Intel architecture. It was the first of its kind and TÜV-certified. No one else offers a certified safety robot controller like this. It gave us a strong base and experience.
As human-robot collaboration increases, safety becomes critical. These applications need both edge AI and safety systems. We have already delivered customised safety solutions for several humanoid robot companies. It is complex and time-consuming—few companies can do it as quickly or effectively.
What are the crucial safety considerations around humanoid robots?
Dr. Ooi: Humanoid robots are mobile and battery-powered. In home settings, with children around, it is crucial they do not collide with people. Safety also comes into play when the battery is running low. The key issue is trust. You do not want a "Terminator" scenario in your house – you must be able to shut the robot down safely anytime. We are also working with NVIDIA on projects in industrial and humanoid robotics in the US, though many are under NDA.
NEXCOM is based in Taiwan but sells in the US, Europe, and other parts of Asia. How do you evaluate Taiwan’s efforts to become a regional AI hub?
Mr. Lin: Taiwan is at the centre of global computing with a strong tech base. NVIDIA recently announced its expansion here with the NVIDIA Constellation, which will draw significant resources. The government is investing around US$300 million to support robotics start-ups and promote AI in five areas: computing power, data access, funding, talent, and marketing.
They are also improving immigration and visa processes to attract skilled workers. Taiwan’s TSMC success was government-driven, and they are trying to replicate that in robotics. With many component manufacturers already here, the goal is to unite them to build general-purpose robots—creating a contract manufacturing model for robotics, like TSMC does in semiconductors.
You described a future with no boundaries for factories—a fully interoperable, global manufacturing system. How do you envisage this in practical terms?
Mr. Lin: Since 2015, we have worked with Azure and AWS, using their cloud and virtualisation tools. By 2016, we saw that having mega factories in each region made more sense—otherwise, logistics are too costly and supply chain risks are unpredictable. Localized manufacturing systems surrounded by supplier ecosystems are agile and resilient in responding to critical crises. Companies like Mercedes-Benz and Audi manufacture in the US, Mexico, and China.
In the future, more companies will adopt distributed factories as regional manufacturing centers. But to manage them, everything must be digitalised and centrally controlled via managing SaaS by hybrid cloud. That is the “boundless factory”(or no borders factory) concept. We offer the NexDATA AI Agents, our LLM-based knowledge and manufacturing management systems to make smart manufacturing a reality.
As rules for reporting carbon emissions become increasingly stringent, what hardware or software solutions does NEXCOM offer that enhance oversight in that domain?
Dr. Ooi: We developed NextDATA ESG, which calculates the carbon emissions generated during our production. By connecting directly to each piece of equipment at every workstation, we collect accurate, piece by piece energy consumption information for individual manufacturing parts.
This is increasingly important as ESG regulations tighten. Companies need to understand their carbon output and how much credit they need to buy. This solution is especially useful in Taiwan, where manufacturers need reliable tools to track and report emissions.
