April 10th, 2025
3 min read
Telecom industry veteran Caroline Chan is responsible for Intel’s global network infrastructure strategy and solution development related to 5th-generation wireless technology, including verticals enabled by 5G, AI and edge compute.
In her role, she sees how AI is moving to the edge of the network where decisions can be made instantly. Not only does this speed up manufacturing processes but it also improves data security.
We spoke to Chan, Vice President Networking and Edge Computing, Intel, about how AI is reshaping the network globally, especially at the edge.
Bridges and Barriers
Traditionally, networks were principally optimized for download and communication. That began to change with wireless networks like 4G and 5G, and especially now with Private 5G, as processing can now be done at the edge of the network. We’re at an “inflection point,” Chan says, that is driving innovation in manufacturing and other sectors.
It’s also leading to more demand for upload capacity on the network to share data back to the cloud and data centers from IoT devices and sensors – a challenge AI can help to solve. AI will improve network efficiency with compute analyzing the uplink and downlink activity called “AI for network.” Using AI at the edge, in particular with Small Language Models (SLMs), can dynamically run the network and provide significant energy savings. At Intel, Chan notes, chip designers have put AI instruction natively into the silicon of the latest CPU to significantly improve AI performance at the edge.
However, AI still faces barriers to its widespread adoption, and Chan believes one of the greatest issues is the need to democratize training models. She says AI tools must become more accessible and simplified so businesses of all sizes can use them without needing large, specialized teams.
Chan shares the example of a manufacturing plant in Thailand, where AI is used to streamline the process of shift transitions and enhance productivity using voice technology. In the past, supervisors would spend two hours each shift inspecting the factory floor and ensuring checklists were completed. With AI-powered speech conversion, available in Thai, supervisors can now more easily communicate essential updates to the next shift, reducing the time spent on this task significantly. This use of AI not only boosts productivity by 10 to 15 percent but also democratizes access to technology, improves productivity, and supports more robust training of AI models for a specific industry use case and need.
AI in Society
I cannot emphasize enough how much collaboration we need between industries, companies and across geographies especially now with the need to power the edge with AI. We cannot just design for the United States, we must design technology for compute and networking for the world.
Caroline Chan, VP of Networking and Edge Computing, Intel
Standards bodies are critical to ensuring interoperability across and within the network globally – AI has made this even more important. In addition, Chan says, the transport layer has to fundamentally change, as does silicon design, in order to enable AI development and adoption. With this in mind, Intel and NTT’s collaboration on the IOWN (Innovative Optics and Wireless Network) Global Forum will support the development of technologies for rich, powerful and energy efficient compute and networking at the edge.
Chan highlights how collaboration between tech companies – such as NTT and Intel – is crucial to achieving the goal of global interoperability for efficient compute and networking. Such partnerships focus on sharing knowledge, establishing standards, and creating the infrastructure necessary for AI to thrive, for the good of all.
“It is essential for us to partner together, for any business to partner, because AI creates such a bold and bigger future, no one company can provide it all,” says Chan.