2023 kicked off an arms race in artificial intelligence (AI). Big tech companies spent heavily on software development and specific chips from companies like Nvidia to train their AI models. As a result, investors saw a few select chipmakers emerge as big winners as demand from deep-pocketed tech innovators drove sales significantly higher.
But the next class of AI winners won't be the companies producing chips to train AI models; it'll be the companies producing chips that use those AI models to produce responses to prompts. Demand for these so-called inference chips will increase as more people demand the ability to use AI on their devices instead of sending data to a server and waiting for a response. Analysts at UBS see inference chips moving from 10% of all AI chips sold in 2023 to 20% by 2025 and $30 billion in total revenue within five to 10 years.
On-device AI is essential for applications that require low latency and reliability (like autonomous vehicles) or where you can't compromise on data security and privacy (like medical or government records). As demand for on-device AI increases, several companies stand to benefit.
Here are three that could be surprise AI winners in 2024.
Image source: Getty Images.
1. Qualcomm
Qualcomm (QCOM 1.45%) CEO Cristiano Amon told analysts in November: "We expect high-performance on-device AI to become a requirement over the next few years." And he sees Qualcomm as a leading beneficiary of that trend as it drives demand for more powerful chips and could expand to new device categories.
Qualcomm is working closely with leading AI models from Microsoft, Meta Platforms, and Baidu to ensure its Snapdragon smartphone chip designs can generate responses faster and more power-efficiently than competing chipmakers'. Meanwhile, it sees opportunities to take share in PCs while helping bring the headset-computer format (virtual or augmented reality) to a broader audience. It also plays a key role in automotive computing.
In the meantime, Qualcomm has a very strong position in the smartphone industry already. Its modem designs are ubiquitous, and its mobile chips are featured in many high-end Android devices. While Qualcomm will lose Apple's licensing business at some point in the near future when it shifts to its own modem designs, it should be able to recover quickly as its position with most Android handset makers is secure and it expands into new device categories.
As demand for power-efficient, high-end chips capable of on-device AI inference grows, more and more device makers will turn to the higher-end chips Qualcomm makes. With shares trading at just 15 times analysts' estimates for this year's earnings, the stock appears fairly priced considering it should be able to produce steady revenue growth from its core business with high upside as on-device AI inference takes off.
2. MediaTek
MediaTek (2454) is Asia's largest chip-design house. Its biggest focus has been multimedia and wireless technologies, but it's expanded to mobile devices and Internet-of-Things products. Last August, the company signed a deal with Meta to build a "complete edge computing ecosystem designed to accelerate AI application development on smartphones, IoT, vehicles, smart home, and other edge devices."
The big push by MediaTek into smartphones and AI chips represents a major threat to Qualcomm, and the Taiwanese company's already made some major wins over the bigger chip designer. Meta chose MediaTek to power its augmented reality glasses in November.
The new Dimensity 9300 chip design is positioned to compete directly with Qualcomm's highest-end chip, which could help MediaTek expand its presence into more flagship devices.
More promising, however, is the Dimensity 8300, which is designed for midrange smartphones. As the first midrange smartphone chip designed for on-device generative AI, it could come to dominate an increasingly important market.
With shares trading at just 16 times 2024 earnings estimates, the stock appears to be trading at a great value considering the potential growth from its push into high-end smartphones and cementing its position in midrange devices.
That said, investors will have to buy shares of MediaTek stock on the Taiwan Stock Exchange, since that's the only place its shares trade. That requires a broker that supports foreign stock trades as well as the ability to stomach fluctuations in foreign exchange rates. On top of that, your broker may charge extra fees for accessing a foreign exchange.
As with any investment in a foreign company, whether it has American depositary receipts (ADR) on a U.S. exchange or trades on a local exchange, there are geopolitical and regulatory risks involved. For example, investing in MediaTek, or any Taiwanese company, is subject to the risk of conflict in the Taiwan Strait between Taiwan and China, or just the perception of that risk (even if the actual risk is low). As such, this is a riskier play on the future of AI and requires investors to keep their eyes on things. Investors with a higher risk tolerance might consider it, but more risk-averse investors might consider the other potential surprise AI winners discussed here.
3. Apple
Apple isn't exactly a big name in generative AI, but that doesn't mean it won't stand to benefit from booming demand for on-device AI inferences. Apple is a leading proponent of data security. It wants to help you keep your personal information on your personal device instead of sending it to a server, and it's making the chips that make it possible to keep everything on its devices while running some of the most advanced AI algorithms built.
Apple builds some of the most advanced smartphone processors on the market for its iPhones. It has a strong relationship with Taiwan Semiconductor Manufacturing Co. to get its hands on the most powerful and power-efficient chip processes available on the market. And its vertically integrated approach puts it in a position to become the only smartphone maker that can offer some of the most advanced AI technologies on its devices. Not only could that help Apple sell more devices, but it could also push more developers to offer new apps in its App Store.
Apple is also reportedly developing its own generative AI capabilities. It could integrate generative AI into iMessage or Siri and instantly have a massive user base, and it could further lock its already loyal customers into its ecosystem. Alternatively, it could charge a subscription fee to access more advanced on-device generative AI features.
Apple trades for a premium price of more than 28 times analysts' earnings estimates for 2024. That said, the stock deserves a slight premium due to its massive net-cash balance and its generous share-repurchase program. While it might not be the typical AI play, it certainly stands to be a big winner as more AI processing moves on to its devices.
