This Is AI’s Most Underrated Opportunity, And The Eight Stocks At The Center Of It

This Is AI’s Most Underrated Opportunity, And The Eight Stocks At The Center Of It
Stéphane Renevier, CFA

5 months ago6 mins

  • End-device AI has notable advantages: it offers quick decision-making, instant access, enhanced privacy, and reduced cloud server load. And it’s going to be everywhere.

  • The opportunity is big for chipmakers, driven by hefty demand from smartphones and tablets, the Internet of Things, and automotive stuff.

  • Bank of America says TSMC, Qualcomm, MediaTek, Nordic, STMicroelectronics, Infineon, SK Hynix, and Renesas are poised to benefit most.

End-device AI has notable advantages: it offers quick decision-making, instant access, enhanced privacy, and reduced cloud server load. And it’s going to be everywhere.

The opportunity is big for chipmakers, driven by hefty demand from smartphones and tablets, the Internet of Things, and automotive stuff.

Bank of America says TSMC, Qualcomm, MediaTek, Nordic, STMicroelectronics, Infineon, SK Hynix, and Renesas are poised to benefit most.

Mentioned in story

When people talk about AI investment opportunities, it’s usually all about the cloud. But Bank of America’s been looking at an AI investing theme that’s a lot closer to the ground – and poised to be just as big. It’s end-device AI. Let’s take a look at what it is, why it’s exciting, and the eight stocks that are poised to benefit from its boom.

What’s end-device AI?

Think of this as a deviation from the usual server-based or cloud-based systems. Basically, every single day, a multitude of devices are spewing out vast amounts of data, and sending it to one of three places to be processed:

The cloud. OK, you’re familiar with this one. All that data goes to powerful yet remote servers. For instance, when you ask Alexa a question, your voice is sent far off to Amazon's servers, processed, and then the answer is sent back to you. This is the most common method for complex generative AI tasks, like running ChatGPT.

The “edge”. The data gets ferried off to nearby locations like offices, 5G towers, or some other physical location, and is stored closer to where it’s generated. For instance, in a smart factory, sensors might gather data about machinery, and then a server within the factory might immediately analyze that data to make real-time decisions, without needing to send it to a remote cloud server.

The device itself. This is end-device AI. All the processing magic happens right on your gadget – be it a smartphone, a smartwatch, or a car. We've seen this speedy processing action with Apple’s new "on-device dictation" features on the iPhone, where some voice processing can occur right on the device without an internet connection.

How data can be processed. Source: Bank of America.
How data can be processed. Source: Bank of America.

OK, so what’s so exciting about this?

We all know the AI revolution is huge. And end-device AI is going to be massive within AI, with its notable advantages:

  • Speedy decisions. Tasks are handled on the spot, so: faster response times.
  • Always ready. You get immediate, easily accessed AI functionalities, because the device is already in your hands.
  • Better privacy. Local processing means sensitive info (think: facial and fingerprint scans) is more secure.
  • Less strain on cloud infrastructures. With devices tackling AI tasks, that eases pressure on distant cloud servers, optimizing overall system efficiency.

And though end-device AI has been very much under the radar amid all the AI hype, it’s poised to explode onto the scene – and soon. System on a chip (SoC) is a kind of mini-computer that houses everything from a device’s core brainpower to its memory and visuals. Right now, some firms are already incorporating dedicated AI functionalities directly into these SoCs. This means devices can process AI tasks with impressive speed, locally, without relying on remote servers. Due to their extensive usage, consistent engagement, and significant data-gathering abilities, smartphones and tablets stand out as the main recipients of these AI enhancements at the device level. But end-device AI's potential extends far beyond smartphones and tablets.

In the swiftly changing realm of the Internet of Things (IoT), everyday items, from refrigerators to smart TVs, are going online and becoming smarter, with the ability to learn and make choices. In the auto industry, AI is bolstering safety and elevating the driving experience. With the momentum toward self-driving cars, built-in AI is increasingly vital.

How big is the opportunity?

Bank of America estimates that the total addressable market (TAM) for end-device AI will be immense: about $5.1 billion for foundries (the companies actually manufacturing chips) and $14.4 billion for chipmakers (the companies designing – and sometimes manufacturing – chips) in 2025, representing 96% and 105% average yearly growth until 2025, respectively.

For chipmakers, the 2025 TAM projections break down like this: $3.1 billion for assistive AI on smartphones and tablets, $6.7 billion for generative AI on smartphones and tablets, $3.7 billion for AI Internet of Things, and $1 billion for automotive stuff.

It’s a lot of growth, but it’s driven by two big factors:

First, there’s the need for stronger hardware as AI evolves. Take smartphones, for example: as AI tasks become more advanced, the existing AI components in the central chip might struggle to keep up. That’ll drive demand for better, and more specialized AI chips.

Second, there's the allure of better tech, i.e. "replacement demand". Imagine the next generation of phones equipped with cutting-edge AI features, able to perform all kinds of impressive tasks. This could compel a lot of people to upgrade, driving a surge in smartphone sales – and demand for end-device AI.

Mind you, end-device AI does have its own set of challenges. For starters, all those added features mean balancing enhanced capabilities with battery life. There's also the financial component: as AI abilities increase, so will the cost of the integrated semiconductors. Plus, the complex algorithms that drive AI and machine learning must be meticulously refined to function seamlessly within smaller devices. And not to be overlooked is the heightened need for cybersecurity: as our devices get smarter, they also need to be fortified against increasingly sophisticated threats.

So, how can you benefit?

Bank of America says these eight companies right now are poised to be the biggest beneficiaries of the coming end-device AI boom:

TSMC: Renowned for its advanced chip manufacturing, TSMC operates as a foundry, producing chips for other companies instead of designing its own. With its technological prowess, vast capacity, and wide client base, TSMC stands poised to benefit significantly from the growing demand for end-device AI chips.

Qualcomm: Operating as a "fabless" company, Qualcomm designs semiconductor chips instead of producing them. It’s a global frontrunner in smartphone chipsets and pioneer in infusing AI into smartphones and edge devices. Additionally, it’s broadening its horizons into the IoT and advanced vehicle tech.

MediaTek: Another "fabless" company, MediaTek designs chips for a wide range of devices and is making big moves in AI. With big-name partners like Nvidia and a leading position in the smartphone/IoT market, it’s poised to lead in the end-device AI sector.

Nordic: Nordic Semiconductor, based in Norway, makes chips for devices that connect wirelessly, like smartwatches. It specializes in a type of wireless connection called Bluetooth Low Energy and is exploring other wireless technologies too.

STMicroelectronics: This worldwide chip maker serves various industries like cars, phones, and networks. It produces power-related products for AI uses, from general computing chips to specific AI-enhancing components.

Infineon: A leading semiconductor company known for its strong presence in the automotive, power, and smartcard sectors, it works closely with major car manufacturers, tech companies, and payment solution providers.

SK Hynix: It’s the No. 1 supplier of a special type of memory called “high bandwidth”, which is essential for Nvidia’s popular AI training devices.

Renesas: It focuses on two things: endpoint AI that completes information processing on terminals, and edge AI that processes information within a closed network. To reduce power consumption, which is an issue for endpoint AI and edge AI, Renesas is also developing a unique AI accelerator.

If you’re thinking of seizing the opportunity, consider diversifying your investments across various companies. And remember, only invest what you’re willing to part with – like much of tech, AI investments can be a rollercoaster with potential big drops.

Finimize

BECOME A SMARTER INVESTOR

All the daily investing news and insights you need in one subscription.

Disclaimer: These articles are provided for information purposes only. Occasionally, an opinion about whether to buy or sell a specific investment may be provided. The content is not intended to be a personal recommendation to buy or sell any financial instrument or product, or to adopt any investment strategy as it is not provided based on an assessment of your investing knowledge and experience, your financial situation or your investment objectives. The value of your investments, and the income derived from them, may go down as well as up. You may not get back all the money that you invest. The investments referred to in this article may not be suitable for all investors, and if in doubt, an investor should seek advice from a qualified investment advisor.

/3 Your free quarterly content is about to expire. Uncover the biggest trends and opportunities. Subscribe now for 50%. Cancel anytime.

Finimize
© Finimize Ltd. 2023. 10328011. 280 Bishopsgate, London, EC2M 4AG