What is MediaTek's footprint in the global IoT landscape?
MediaTek is one of the top five global semiconductor fabless companies, alongside giants like NVIDIA, Qualcomm, Broadcom, and AMD. We ship around 2 billion devices annually, and are number one in several sectors, including smartphones by volume, smart TVs, home broadband Wi-Fi, Arm-based Chromebooks, and Android tablets. Our focus is now on automotive, IoT, and data centers as emerging opportunities. At the core of our success is the integration of intellectual properties (IPs) into system-on-chips (SoCs), leveraging leading-edge process technologies in collaboration with partners like TSMC and Intel Foundry. We also pride ourselves on our pace of execution, offering products at competitive prices, which enables scale-based adoption.
In the IoT space, the market is highly fragmented, with no clear top players. IoT tends to leverage existing products designed for other markets, enhancing them to meet specific IoT requirements, such as ruggedization for automotive or extended temperature support for industrial applications. We invest around $4 billion in R&D annually to adapt and repackage existing products for diverse markets like smart fitness, retail, industrial automation, transportation, and smart cities, to meet the varied needs of IoT customers without needing custom-built solutions for each market.
In March, MediaTek launched the Genio 720 chips, with 250% enhanced AI capacity on its predecessors. Can you contextualize, for a non-specialist audience, why this is significant?
The need for enhanced chip capacity, particularly in AI, is driven by the growing demand for edge computing.
AI at the edge reduces latency, which is essential for applications like robotics or autonomous driving where real-time responses are critical. The key is to perform processing locally rather than relying on cloud-based systems, which introduce delays.
From an economic standpoint, continuously sending data to the cloud can be inefficient, especially for large amounts of data like 24 hours of traffic footage. The Genio 720 chip enables smarter, more efficient edge computing by distilling large language models and AI tasks to fit the capabilities of edge devices, making them more responsive without overwhelming cloud resources.
The significance of the Genio 720 lies in its ability to support distilled versions of large language models, which are scaled-down for edge applications. This makes it feasible for industries like industrial automation or smart homes to interact with devices using natural language without needing massive cloud processing power. Our platform offers pre-built frameworks, making it easier for customers to implement AI solutions without needing a team of AI engineers, and providing a ready-to-use solution that can be customized for specific vertical applications.
Do you believe enough tech company energy is going into solving power efficiency alongside AI functionality?
Power efficiency has always been a core focus for MediaTek, especially given our background in smartphone technology, where battery life is crucial. Every decision, from selecting IP to choosing memory, is made with power efficiency in mind. In contrast to companies focused on raw performance, we have always prioritized maximizing the value of each milliwatt to extend usage time. This mindset is now being applied to edge computing, where power-efficient computing plays a key role in sustainability. By reducing the need for data to travel back and forth to the cloud, we save not only energy but also enhance privacy by keeping more processing local.
I believe that power efficiency is becoming a more significant focus across the tech industry, but it is still an area where more attention is needed. Edge computing plays a critical role in reducing energy consumption at the aggregate level, as it minimizes unnecessary data transfers. By shifting from large, inefficient cloud-based solutions to more localized, intelligent edge processing, we can make a meaningful impact on overall energy usage and contribute to more sustainable tech infrastructure.
What is your prognosis for the massive investment in data centers, now that there is an accelerating shift towards hybrid and local processing and away from reliance on cloud-based technologies?
I believe it is not going to be a matter of one or the other, but rather both. Cloud computing will continue to grow due to its massive capacity and ability to process large amounts of data. However, we foresee edge computing growing at a faster rate. We are reaching a tipping point where the capabilities previously only available in server-class hardware are now being integrated into small, fanless devices. For instance, we are working on building the world’s smallest audio system-on-module (SOM), which demonstrates how compact and efficient edge technology is becoming. Disruptions like this will accelerate edge computing, helping it catch up with the growth seen in data centers.
As edge computing continues to grow, it will coexist with cloud infrastructure, with each complementing the other. The cloud will remain essential for certain tasks, but edge computing's ability to process data locally will drive innovation in areas that require real-time responses and low latency.
What has impressed you in the last few months in terms of technological advancements in consumer devices and technology?
What has truly impressed me recently is the creativity surrounding the use of generative AI at the edge. The potential applications of AI in our daily lives—whether it is in how we work, live, or play—are fascinating. The creativity unleashed by this technology, such as on-demand video and image generation and its ability to enhance human connections, is phenomenal.
At a broader level, the challenge is that technology’s benefits are not always visible, as people often associate it with the negative side effects like teenage depression and social media issues. But technology is constantly improving every aspect of our lives, even if we do not always notice it. As engineers, we need to do a better job of highlighting these benefits and ensuring that technology serves society in a way that minimizes harm and maximizes improvement. I think we can take a more anthropological approach to technology, with the aim of it demanding less interaction from us, and instead fading more into the background so we do not spend so much time dealing with technology to optimize our lives.
