After the US announced more stringent curbs on exports of advanced artificial intelligence (AI) chips to China, the darling of the Standard & Poor’s (S&P) 500 — Nvidia Corp — saw its shares tumbling almost 9% in the last week.
The chip manufacturer’s two new chips, the A800 and H800 chips, designed specifically to sell to China under AI chip restrictions announced last year, became subject to the new regulations. As a result, technology-centric ETFs that had heavy exposure to Nvidia also suffered from the drag. Despite the recent decline, Nvidia’s share price has tripled since the generative AI boom started late last year.
Evidently, investors chasing Nvidia fiercely all the way to its recent peak were not the only ones surprised by the AI boom. According to Dominic Rizzo of T Rowe Price, the leading mega-cap technology companies were taken by surprise by the magnitude of consumers’ response to ChatGPT, the popular AI chatbot that has become the shorthand for the current AI boom.
The result has been what he describes as an “arms race” to acquire new capabilities and refine existing ones. Rizzo, who is the portfolio manager for the global technology equity strategy at T Rowe Price, therefore believes that generative AI will reshape the investment landscape as we know it.
But amid the volatility in the stock market caused by geopolitical tensions, how should investors invest in AI? Rizzo, who was first put in charge of semiconductor investments for the firm in 2015, employs a four-pronged approach to picking out opportunities in different regions, sub-sectors and market capitalisations.
“A wide range of companies, both large and small, stand to benefit from development of AI applications,” he says. “And in our view, companies of nearly any kind — and their investors — need to pay attention.”
Rizzo: A wide range of companies, both large and small, stand to benefit from development of AI applications. Photo: T.Rowe Price
First, take note of companies that sell mission-critical “linchpin” technologies. The analyst defines this as companies that have technologies that are so crucial to the overall ecosystem that if they were to disappear tomorrow, the lives of the everyday man would be dramatically worse off.
Next, these companies should be innovating in a secular growth market, in which they should demonstrate taking shares quickly, and growing fast. These companies must also show improving fundamentals — accelerating revenues, high incremental operating margins that are expanding, and free cash flow conversion that is improving. This shows that they are making efficient use of every dollar of revenue.
Finally, companies need to have “reasonable” valuations. Rizzo says that investors get burned when they buy incredibly expensive stocks, but on the flip side, do not see much value in investing in cheap stocks. By considering all four factors before investing, the analyst believes that there will be opportunities everywhere.
“If we do that, I think we can find opportunities in all geographies of the world, from the US to Europe and Asia; in all different sub-sectors of technology hardware, software and Internet payments; and of all different market capitalisation rates, ranging from small cap to mega cap. That’s what we’re trying to do,” he says.
Where are we now in the AI cycle?
The analysts at T Rowe Price have long been following and investing in the basket of interrelated innovations that had to come together to build ChatGPT and other “foundation models”.
These include cloud computing, new means of efficient communication between computing systems through application programming interfaces (APIs), and the accumulation of sheer computing power enabled by ever‑faster chips and processors.
For investors, Rizzo says it is crucial to understand the background and scale of generative AI.
While AI has been around for a long time, what makes generative AI powerful is its ability to create. The analyst likens it to a masterchef who is able to cook, generate their own recipes and generate new techniques by themselves.
At present, the world is at a centralised computing stage, building out huge graphics processing unit (GPU) clusters at hyper scalers, also known as large cloud service providers. In order to train these large language models (LLMs), we are going to move to the decentralised computing stage, which is going to put AI into our smartphones, Rizzo says.
These LLMs can be applied to chat generation, cybersecurity threat detection — virtually every sector of the global economy, the analyst adds. For the technology sector, this would immediately mean a boom in the chip market, as training LLMs is a silicon-intensive activity.
“The thing that excites me most is the general silicon intensity of AI,” he says. “For example, an AI server can consume up to three times more DRAM [dynamic random access memory] than a traditional server and eight times more storage.”
Semiconductor companies and potentially hyper scalers are going to try to design their own chips, Rizzo predicts. Advanced Micro Devices (AMD), another leading chip player, projects that the AI chip market will grow from US$30 billion ($41 billion) in 2023 to US$150 billion in 2027, representing almost a 50% CAGR, Rizzo notes.
But just as Nvidia’s AI chip is behind ChatGPT, so are other select companies behind computer chips labelled Nvidia, AMD or Intel. A crucial one, says Rizzo, is Taiwan Semiconductor Manufacturing Corporation (TSMC), which has become a widely recognised leader in making ever‑smaller process nodes — a measure of how finely transistors can be etched onto silicon and the number of transistors can fit on a chip of a given size.
How to enable and implement AI going forward?
Rizzo describes four layers of AI companies in understanding investing in different stages of AI. At the bottom are chip ecosystem enablers. This includes Nvidia, ASML, Samsung, TSMC among others.
The level above that are infrastructure enablers who will enable the implementation of this technology. This includes Amazon Web Services (AWS), Microsoft Azure Cloud, Alibaba Cloud and others like MongoDB.
Next are the foundational players. The most important LLM that everyone has heard of, says Rizzo, is ChatGPT, but others such as Google’s PaLM and Amazon’s Triton are important in encouraging innovation over time.
The final layer at the top, which Rizzo describes as the hardest bit to understand at the moment, is the application layer. “Trying to call the winners at the application layer today would be like trying to call the fact that Uber would be pervasive all over the world,” Rizzo says. “We don’t know who’s going to win at the application layer.”
What does it take to win at AI?
The business potential of AI is enormous, but before rewards can be reaped, there are some very costly investments that have to be made. The team at T Rowe Price have four pointers to identify winners — compute resources and capital; data; distribution and talent.
Using Microsoft as an example, Rizzo notes that the company’s capital expenditure budget was forecasted to go up sequentially from each quarter here on. In addition to that, a large amount of data sets is necessary to “rip apart” and eventually obtain insights, he says.
Next, firms need to have a large enough customer base to push their AI innovations to, and there should be stickiness among consumers to fund the continuous innovation.
Lastly, only a few people in the world have the technical expertise to work in AI, in which many have been acquired by major technology companies in the US, and hyper scalers alike.
For this reason, Rizzo is in the camp that AI is a sustaining innovation. “If you are a winner, you will continue to be a winner,” he says. “Because you’re the type of company now that has computing resources and capital, distribution, talent, and data. I don’t think AI is a disruptive innovation; it’s a sustaining one that takes these big companies giving more power.”
On that note, the analyst believes that AI at the end of the day will be a “huge productivity enhancer”, in which the global economy will be able to implement these technologies into various software applications and sub-sectors.
Finally, on the back of the ensuing US-China chip war, Rizzo sees both markets developing differently. He likens it to the development of the Internet in the two regions, where the winners in the US are the likes of Google and Amazon, while the winners in China are Tencent and Alibaba among others.
“But this is still really new to the US Congress, department restrictions came out a few days ago,” Rizzo notes. “So I think even the companies are still trying to figure out exactly how these new restrictions are going to affect the market, and our job is to navigate that environment responsibly.”
