Our first Ubuntu IoT Day in Southeast Asia – and our first ever event in Singapore! It was long overdue, as several attendees were quick to remind us. Ubuntu has long been a quiet force in the region, powered by its rich ecosystem of innovators. More than 150 participants came together to represent Southeast Asia’s diversity of thought and culture; to challenge, learn, and discuss the future of IoT.
The event took place on May 27th, and it featured 9 presentations from leading industry experts like ARM, Intel, Qualcomm, MediaTek, Advantech, and Aaeon, discussing topics like edge AI deployment challenges, trends in robotics, regulatory compliance, as well as hardware and software tools for the IoT sector. In this blog, we’ll dive into the three key topics covered in these sessions.
From training to deployment: the challenge of AI at the edge
A theme that quickly arose was how to manage the challenge of deploying AI at the edge. Several presenters highlighted the complexities companies face: having the right data, selecting the right model, tuning it for performance, validating it on-device, and managing updates at scale.
Qualcomm’s presentation addressed these points head-on, introducing their AI Hub – a repository of pre-optimised models designed to run efficiently on Qualcomm chipsets. The solution appeals to developers and product teams in sectors like automotive, robotics, and smart devices who work with Qualcomm hardware and need high-performance AI at the edge deployments. Their approach emphasises simplifying deployment pipelines and reducing time-to-market, particularly in industries where real-time performance matters.
Intel also participated in this theme, noting that despite high interest, only a small fraction of edge AI projects actually reach production. The reasons vary – from lack of clean data to the complexities of hardware-software integration – but the result is the same: delayed innovation. Intel’s solution is to accelerate deployment through Verified Reference Blueprints, providing a clear, validated stack that includes hardware configurations, memory, storage, and the software layers needed to bring AI to life at the edge. These blueprints could be especially valuable for solution architects and engineering teams looking to streamline proof-of-concept to production transitions.
MediaTek, too, contributed to this conversation by showcasing how their Genio platform uses Snap packages to securely deliver AI applications. By leveraging Canonical’s Snap technology, they enable modular and transactional updates – ensuring devices stay current, without disrupting operations. This approach is particularly relevant for OEMs and device manufacturers aiming to maintain secure, updatable AI capabilities across diverse product lines.
Software defining hardware – the new reality for Silicon and ODMs
Silicon and ODMs, traditionally hardware-first companies, today are operating in an increasingly software-first world. ARM, Intel, Qualcomm, MediaTek, and Aaeon all echoed this sentiment in different ways. Hardware must now come with software ecosystems that allow developers to quickly build, test, and deploy. Otherwise, their current offering fails to adequately meet the needs and usage requirements of their products.
ARM, for example, presented Kleidi—a middleware layer that abstracts its hardware complexities, allowing AI engineers to run models using standard frameworks like PyTorch. With Kleidi, ARM is aiming for more models to be deployed directly on CPUs, removing the need for proprietary GPU SDKs and licensing constraints. This is especially compelling for AI developers and platform vendors seeking greater portability, reduced costs, and streamlined deployment across ARM-based devices.
Their message was clear: they no longer see themselves purely as a hardware provider, but as a middleware company, delivering integrated, secure solutions for enterprise edge environments.
How to ensure secure deployments and meet compliance
The event was also packed with sessions covering security compliance and tooling. These topics were enthusiastically welcomed by the audience – we saw the same reactions during our IoT London day last year.
Compliance, particularly in the context of the European Cyber Resilience Act (CRA), was a hot topic. Attendees expressed strong interest in what the CRA means for global markets. While the CRA is an EU regulation, many vendors recognize its global implications. Advantech, for example, is already aligning with IEC 62443 standards and is confident that CRA-like regulation will soon come to Asia. The general consensus was that regulatory pressure will only grow, and companies that embed secure development practices now will have a significant advantage.
Canonical presented a deep dive into the CRA. You can find more information about the implications for device manufacturers in our blog post. Canonical is dedicated to fulfilling the requirements of the CRA. This includes delivering compliant software, conducting attestation for our non-critical software, and adhering to manufacturers’ specifications for the software our customers obtain through us.
Advantech also showcased its DeviceON platform –a remote management tool that helps enterprises monitor and update fleets of devices in the field. The solution seems particularly valuable for system integrators and IoT solution providers looking to simplify large-scale device deployment and maintenance. Advantech highlighted Ubuntu Pro as a key component in their security story, bringing security maintenance for more than 36K packages, for up to 15 years.
‘Til next time
It’s always a privilege to engage with our Ubuntu users, especially when we can host them. Our IoT days aim to deliver an event that expands the technical knowledge of our participants and allows them to grow their professional network through our partners.
Singapore might be our first stop in Southeast Asia, but it definitely won’t be the last.