For more than 43 years now, Advantech has been a global leader in the design and manufacture of embedded and ruggedized computing systems deployed in a broad range of tech verticals ranging from healthcare and retail to industrial manufacturing and processing.

Today, the company finds itself at the confluence of rapid innovation on the AI front and the technology's move from cloud-based crunching of large language models into the physical world of robots and other locally controlled machines. As such, the company suitably added an Edge AI Conference to its annual World Partner Conference, held the week of June 1st in the company's headquarters city of Taipei, Taiwan. A broad range of the company's tech partners, OEMs, customers (and yes, journalists such as yours truly) were on hand to share and learn more about the latest developments shaping this new frontier—and Advantech's plans to help coax it into reality.

While Advantech doesn't design or manufacture the silicon chips that make computing and edge AI possible, it has developed a range of software tools for creating and managing the applications and models that run on them. Advantech WISE is the company's industrial AI platform and solution portfolio that enables edge AI across machines, systems, and sites—turning real-time data into intelligent actions across diverse hardware environments.

In March, 2026, Advantech added WEDA (WISE-Edge Developer Architecture) to its portfolio. This framework, Advantech CEO KC Liu noted in his keynote address, is available to the company's customers and developers free of charge, and helps to solve a range of challenges starting with deployment of software models across a range of different chipmakers' platforms. Further, it is designed to help manage deployed devices at scale, whether than means dozens, hundreds or even thousands of devices. The framework also unifies data from diverse sources, simplifies hardware integration, and allows for proactive monitoring and continuous improvement.

The edge: AI's next frontier

While data centers continue their explosive growth, the market for edge AI is poised to grow robustly as well, noted Miller Chang, president of embedded sector for Advantech. "The U.S. market alone for edge AI is projected to grow to $197 billion by 2034," he said.

The edge has great advantages over the cloud and enables physical AI, Chang explained, referencing the many diverse robotics solutions already proliferating in the market and the more than 50 global providers of robots, microprocessors, applications and related solutions that Advantech already counts as partners.

"As AI enters the physical world, the business opportunity will be several orders of magnitude larger," agreed Deepu Talla, vice president, robotics and edge computing for Advantech partner Nvidia. These systems, which are based not on likely word order but on the physics and molecules that surround us, are poised to proliferate, Talla said, but some issues persist.

Of central importance are functional models of the real world, such as those used to simulate the performance of semiconductor chips, each of which may contain more than 100 billion transistors. Nvidia's simulations allow it to assure the performance of a new GPU design--before one is ever manufactured. "AI models need this sort of accuracy, but we haven't yet fully cracked that code [for physical AI]," Talla said. Plus, they need to be integrated with existing workflows in the real world. "Robots need to play nice with all these pre-existing systems," he said.

"Until these problems are solved, physical AI will not take off," Talla predicted. "Deploying robots is only the start of the journey, a continuous lifecycle journey."

Another major constraint in the industrial space is the persistent disconnect between operational and information technology architectures, said Praveen Tamang, global head of automation software, Bosch Rexroth. "It's a huge gap," Tamang said, "and many pilots never scale because of monolithic, legacy architecture."

Data center intelligence for the real world

System-level expertise is also making next-gen robotics a reality, added Nakul Duggai, Qualcomm executive vice president and group general manager for automation, industrial and embedded IoT and robotics. "While the rate of change across industries varies, we're seeing a tremendous amount of change across them all," he said. "We serve industries with a relatively general approach, putting key component systems into the hands of developers," he added. "The use cases will continue to get more and more complex, and robotics will continue to advance."

"We're talking about extending intelligence into the real world," added Ed Dorran, PhD, vice president of strategy for the Edge AI Foundation, a California-based community of edge AI developers, technology makers and academia that serves as a global hub for energy-efficient AI technologies. "Edge AI has a vast ecosystem that is poised to take data center intelligence into the real world," Dorran said. "But edge AI systems need to sip energy, communicate over very low bandwidth and still deliver value with low cyber vulnerability."

"WEDA is both a tool and an architecture for the AI lifecycle, including data and device management across a range of industries," said Richard Huang, chief software architect, Advantech. "While 800 million people--10% of the planet--are using AI already, 40% of AI projects will fail, because the technology is being applied to pre-existing processes."

Despite these obstacles, edge AI technologies remain the focus of leading minds and deep pockets in the technology space. "I believe everyone understands that AI is changing the world," Advantech's KC Liu added. "We can create a limitless future together."