By Partner & Portfolio Manager, Dr. JoAnne Feeney
Economists regularly plead for politicians to seek efficiency in the pursuit of any fiscal goals, even though those pleas invariably fall on deaf ears. National security is one of those goals that has come into greater focus for the last two administrations as a growing China military threat has come to the fore. Last week, we were reminded of the potential cost of policy intrusions into private markets when the U.S. placed additional limits on the export to China of advanced technology, notably computing chips and the equipment needed to make them. Companies have always had to navigate around the obstacles created by industrial policy. The repeated rounds of trade barriers being raised both by the U.S. and China make this one of those more challenging times. Fortunately, the companies involved find themselves in an environment with unusually strong global demand.
We have highlighted the power and pervasiveness of artificial intelligence (AI) in prior commentaries, although we have not focused on the military applications that lie at the center of the growing trade war between the U.S. and China. For decades, technological enhancements improved military capabilities across the globe, but the leap made possible by artificial intelligence brings greater urgency to the West’s race to maintain its superiority. In addition to the greater power AI affords surveillance and reconnaissance through unmanned and self-directed aerial, ground, and sea vehicles, AI also makes possible autonomous weapon systems (AWS), both stationary and mobile.
The U.S. is attempting to slow China’s implementation of AI-enabled capabilities by denying them access to the most powerful chips used for “training” AI models. Massive servers using tens of thousands of specialized chips solve highly non-linear computational problems to “teach” a program to be able to tell the difference between an innocent and a threat, and to recognize, predict and intercept those threats. Limiting the number and quality of the chips China can obtain will indeed handicap China’s AI toolkit and so preserve the military leadership of the west for a while (and maybe a very long while). The limits the U.S. imposed in October 2022 were thought to be sufficient to rein in China’s capabilities. As the main provider of such chips, Nvidia responded by selling instead a chip with less computational power. Last week’s additional restrictions remove those lagging edge chips from China’s options. Some of these AI models are so complex that they could take years to be solved without the most advanced chips.
The U.S. also placed restrictions on the sale of the equipment needed to manufacture such chips and pressured its allies (such as the Netherlands and Japan) to do the same. Equipment makers in the U.S. and abroad have been lobbying against these rules and for exemptions, but the U.S. view has begun to prevail. Without the most advanced chip-making equipment, China will not be able to make the most advanced chips (until and unless they make their own equipment). This is because the key features on those chips—the transistors that number in the trillions—are built at the 3-nanometer scale and the machines needed to draw and build features that small are made by only a handful of U.S. and European companies. Stop those machines from getting into China, the reasoning goes, and you freeze in place China’s chip making capabilities.
China reacted over the past year to those initial restrictions by stepping up imports of the chips, it was still allowed to buy and increased orders for the most advanced equipment it could find. With those systems, China can still do quite a lot. It can make some very advanced chips, but the best it has remains a couple generations behind the chip Apple makes, for example, for its iPhone. And Apple’s chip is not nearly good enough to run AI training programs. China is instead using older, lagging edge chips from Nvidia and is trying to run its models longer to remain active in the AI game. The Financial Times reported in August that China had placed $5 billion in chip orders in an effort to build as many systems as possible before the door slammed shut. Without the newest chips, and more importantly, without those coming out next year, the year after that, and so on, China will fall further behind.
Notably, China has been unable to get deliveries on all the orders it placed in advance of these new restrictions. And this is not because the suppliers do not want to make those sales, but rather because demand around the world for those chips remains particularly strong. The wait time for orders to be filled, usually around six to eight weeks, is in many cases now more than six to twelve months. Because of the rapid increase in commercial efforts to build and deploy AI platforms around the world, companies like Nvidia have plenty of customer alternatives to China right now and likely for several years to come. And it isn’t just Nvidia. These AI servers require memory chips, interconnect chips, hard drives, and many other components, so we are in a period where we are likely to see sustained demand for a fairly broad group of electronics suppliers.
This relatively new source of demand comes on the heels of a period of weak demand for the chips used in PCs and smartphones that followed the surge in production to meet pandemic-related demand. We should not expect to see all these new AI systems come to fruition this year or next, but rather over many years as new applications appear. We will still see cycles in demand for consumer electronics, autos, and industrial adoption, of course, but the long-term trend is highly supportive of growth in valuations of the leaders in this segment and we are building in exposure to this trend across portfolio strategies at Advisors Capital.
In so doing, though, we are mindful of the law of unintended consequences. Even as the U.S. would very much like to constrain China’s deployment of AI, China, of course, is investing to develop its own chip design and manufacturing capabilities. Not only has China been able to acquire knowledge of the most advanced equipment in use around the world, it has also brought home many engineers who once worked at those western companies. It took decades for ASML, of the Netherlands, to commercialize today’s most advanced piece of the chip-making equipment recipe. It won’t take China as long. Investors must pay careful attention, as always, to changes in the leaders and followers in these technology races even as governments attempt to manipulate the winners and losers through industrial policy.
The foregoing content reflects the opinions of Advisors Capital Management, LLC and is subject to change at any time without notice. Content provided herein is for informational purposes only and should not be used or construed as investment advice or a recommendation regarding the purchase or sale of any security. There is no guarantee that the statements, opinions or forecasts provided herein will prove to be correct. Past performance may not be indicative of future results. Indices are not available for direct investment. Any investor who attempts to mimic the performance of an index would incur fees and expenses which would reduce returns. Securities investing involves risk, including the potential for loss of principal. There is no assurance that any investment plan or strategy will be successful.