In this CEO-to-CEO interview as part of ipXchange’s sponsored coverage of Ambiq’s presence at CES 2024, Guy chats with Fumihide ‘Humi’ Esaka for an unscripted, free-flowing conversation about how Ambiq aims to tackle the issues of the time with their ultra-low-power microcontrollers.
What does ‘low power’ mean in 2024’s wearables market?
To begin, Guy asks Humi what low power means to him, and as Humi explains, low power consumption is fundamental to the future of electronic systems in general. With demand for longer battery life in many applications – including wearables – low power consumption is key to making this a reality.
The alternative route to improvement for consumer products is the ability to do more while keeping the same battery life, whether that be with better Human-Machine Interfaces (HMIs) or adding new functionality to the latest model. This path still requires lowering the power consumption of many aspects of a product.
A key challenge when developing new wearable devices is that consumer expectations strongly dictate what is seen as acceptable in terms of battery life and product functionality. As Humi recalls, when he started working in this field eight years ago, a smart watch that could last a few hours to a day was sufficient to meet market expectations.
In 2024, two weeks of continuous battery life is taken as standard for a similar product, which undoubtedly will have more features and functions than the smart watches of eight years ago. This can only be the result of extreme innovations in energy efficiency within the main processing and circuitry architectures of the product, as well as the battery technology that powers it.
The independent co-evolution of battery technology and Ambiq’s low-power microcontrollers
Guy then asks Humi how this demand for longer battery life affects the relationship between companies like Ambiq and the battery manufacturers themselves, but Humi claims no direct collaboration in these development processes, at least not in his experience. It is a mutual respect between the advances in battery technology and developments in low-power architectures for semiconductor technologies like microcontrollers that has resulted in the evolution towards improved products that meet the demands set out by the market.
Spurred by a question about what low power will look like in 18 months, Humi then outlines the next step in Ambiq’s product line, which further reduces the power consumption with improvements to both the hardware architecture and to the software that runs on it. This includes a movement towards the ultra-low-power Cortex-M55 core architecture but with Ambiq’s SPOT technology (Sub-threshold Power Optimized Technology) remaining fundamental to this process.
For those wanting to learn more about the SPOT platform and how it greatly reduces the power consumption of Ambiq’s microcontrollers, check out the ipXperience interview with Ambiq CTO Scott Hanson.
Where will Ambiq’s technology find itself as the demand for low power consumption increases?
The applications of any technology that ipXchange covers is of key interest to our members, so Guy asks Humi the question, and with connected medical devices on the rise, Humi believes these types of products will soon require the ultra-low power consumption offered by Ambiq’s chips.
Humi also sees strong potential for Ambiq’s chipsets in industrial and automotive applications, particularly for making sure that improvements in electric vehicle battery technology remain dedicated to increasing the range per recharge cycle, rather than for processing additional sensor data. In this case, Ambiq improves the efficiency of the overall vehicle by decreasing the power consumption of systems that are peripheral to the consumer’s main expectations for the product.
The true battle to meet consumer expectations when releasing new battery-operated products
Humi then takes this idea further regarding the contribution of many systems to the overall power consumption of a product and how even low-power circuits can accumulate, resulting in a product that may no longer meet consumer expectations. Longer battery life and less frequent, shorter charging cycles is what consumers have come to expect with each new generation of a flagship product.
But this is not simply want for want’s sake: Humi outlines that a smart watch that can last 3-4 weeks on a single charge means that you can take it on a business trip without having to worry about power drain due to general usage. It is improvements like this which make the latest product worth buying as the technological advancements result in a superior user experience that is explicitly marketable over that of the previous generation.
As Guy summarises, more data and functionality will increase the processing requirements of a wearable design, so a more efficient microcontroller is required to keep a product’s power consumption as low as possible for longest battery life. While Ambiq is not looking to push the boundaries of what a semiconductor-based sensor can measure, for example, it is absolutely dedicated to making sure that designers can get more battery life out of any systems that use such innovations, and thus meet the consumer expectations for overall product innovation.
End-point AI in consumer electronics and the challenges posed in implementing this
One such innovation is Artificial Intelligence (AI), possibly the biggest topic of conversation at CES 2024, at least for ipXchange. With many consumer OEMs trying to implement AI for improved user experiences and product functionality, Humi brings to light the key point that an AI function such as live language translation on a smart watch has big implications for battery life.
With natural language recognition and multilingual translation requiring significant computing power, whether on the edge or in the cloud, this is challenging to implement on battery-operated devices while retaining a battery life that is both usable and meets consumer expectations. Despite the hurdles, Humi is confident that Ambiq will be able to bring this functionality to the mass market thanks to the ultra-low-power architecture of its devices and efficient implementation of AI algorithms running on them; the introduction of the Cortex-M55 core to future product lines will undoubtedly help in this regard.
Interestingly, this flips the initial conversation of low power consumption on its head, as Guy rightly notes. If we begin to live in a world where most implementations of AI in everyday life are not plugged in, battery power starts to be used mainly for continuous processing and interpretation of incoming sensor data. This is in stark contrast to simply powering a device that takes in sensor data, performs a logic operation based on a predetermined threshold or wake-up command, and reverts to a low-power mode to conserve energy.
While it is unclear whether it is the tech space or the consumer that originally sparked the growing interest in AI over the last 18 months, one thing seems clear: AI is here to stay, so product manufacturers must all learn how to use and implement it to their competitive advantage.
The lingering question here is how the introduction of AI workloads might change the operating requirements of Ambiq’s devices – from other interviews at CES 2024, it is clear that many manufacturers have rebuilt devices from the ground up to be able to run AI tasks effectively and efficiently. Humi proposes that Ambiq’s architectures are still suitable for such tasks due to their inherently low power consumption, but it will be the clever implementation and management of the AI algorithms and processes that will be the key to unlocking the full potential of this functionality.
This includes new wearable markets, such as AI-enabled glasses that know when you’ve met someone before. With the help of Augmented Reality (AR) displays, such products could also provide information about this person or display live-translated subtitles as desired with the previously mentioned multilingual smart watch.
While live translation functionality exists within many mobile phones and tablets, there is currently a natural language interpretation gap where the speech of non-native speakers may be misinterpreted due to their accent. Humi is confident that these devices will learn and improve with time, but future wearables must be able to clearly understand both the main user and third-party interactions outside the smartphone ecosystem. This in turn will require such devices to run more complex code than is currently used today, so the need for power efficiency is essential for bringing these ideas to market.
The impact of end-point AI to user privacy, data security, and trust in this emerging technology
As Humi explains, AI at the end point is also key to privacy and data security. Any data that requires processing in the cloud is vulnerable to interception and acquisition by third parties, and this data may be of high sensitivity to the user. This means that data and any processing must be confined to the end device in order to keep it secure.
Any attempt at implementing additional security for device-to-cloud communication is another power expense on the system, so AI inferencing done offline needs to be as efficient as possible to maintain long battery life. This goes hand-in-hand with the promise of secure and private data processing, as well as a user experience that meets or exceeds consumer expectations.
All these criteria must be met in order for consumers to accept AI as part of their daily life. The addition of AI to existing products cannot be allowed to negatively impact on user expectations, and user privacy must be maintained, i.e., the device cannot ever be seen to be spying on its user and sending this data to the cloud.
With this in mind, it is clear that low power consumption is at the heart of making AI a viable and celebrated technology for use in everyday life, and this is where Ambiq’s expertise presents a solution that is truly disruptive in today’s MCU and wireless SoC market.
Learn about Ambiq’s Apollo4 MCU and its wearable-oriented evaluation kit by following the link to the board page below, where you can apply to evaluate this, or any of Ambiq’s technologies, for use in a commercial project.