This is the golden age of investing in artificial intelligence, says Eric Schmidt, who used to run Google, now known as Alphabet. UBS calls AI a top investment theme for the coming decade. So, where are all the attractively priced stocks? Don’t suggest using AI to search for them—I spoke with a guy who’s doing just that, and it’s off to a slow start.
Schmidt says machines will cure our diseases and enrich our lives, and they probably won’t annihilate us Terminator-style. But they might trick us into annihilating each other; he recommends making some modifications while we can.
There’s a 3,000-year-old Chinese game called Go that is so complicated, it has more possible board configurations than there are atoms in the universe. When an Alphabet program called AlphaGo became the best player in the world in 2017, it had benefited from seeing humans play. But it was quickly surpassed by a version called AlphaGo Zero, which was told only the rules, and taught itself creative, and previously unseen, strategies. The game, as Alphabet (ticker: GOOGL) puts it, is “no longer constrained by the limits of human knowledge.”
AI enables machines to behave like humans. A subset called machine learning involves training muscular computers using labeled data, so that they can then make educated guesses, or inferences, about raw data, and get better as they go along. A subset of that, called deep learning, uses artificial neural networks modeled on our brains to cut down on the need for human intervention.
There have been two prior AI research booms that fizzled, with the peak of each roughly marked by a movie about a computer gone rogue. In 1968, it was HAL in 2001: A Space Odyssey, and in 1984 it was Skynet in Terminator. This time around the technology has reached escape velocity, for three reasons. First, the world is awash in valuable data. By 2030, humanity will amass enough bits to fill 610 iPhones, each with 128 gigabytes of storage, for every person on earth. Second, advances in data-center chips allow machines to scour that information for insights.
Third, companies are already profiting. AI powers Google’s search results, Alexa’s speech recognition, and Tesla’s self-driving cars. Netflix uses it to not only recommend movies and shows, but also to tweak data speeds on the fly, and guess which thumbnail picture makes a user most likely to click on a title, and even decide the likeliest recipes for creating new hits.
UBS expects AI revenue to grow by 20% a year to $90 billion by 2025. The good news for S&P 500 index investors is that they are already loaded up on key beneficiaries, including Amazon.com (AMZN), Microsoft (MSFT), Alphabet, and perhaps clearest of all, Nvidia (NVDA), whose chips dominate in the training process for machine learning.
Schmidt, the former Google CEO, says big players have an advantage in their wealth of data, while small ones will benefit from a free flow of capital toward AI start-ups. “These waves come along and everyone gets boosted up,” he says. “Not all of them win, but a few of them win tremendously.”
But UBS says investors are better off looking beyond the biggest AI names to stay clear of regulatory risk. Many smaller names are private, or spoken for, like Xilinx, which is being bought by Advanced Micro Devices (AMD). One chip up-and-comer is Marvell Technology (MRVL), valued at $69 billion. Bank of America recently called it the next $100 billion cloud leader. It trades at more than 43 times next year’s projected free cash flow. Marvell and larger Broadcom (AVGO) help cloud giants create their own application-specific chips. There are exchange-traded funds that focus on AI, sort of. The Global X Robotics & Artificial Intelligence ETF (BOTZ) holds plenty of industrial robot makers, like Switzerland’s ABB (ABB) and Japan’s Fanuc (FANUY).
Schmidt says AI could be 10 to 15 years away from beating the stock market. Chris Natividad, chief investment officer at EquBot, is giving it a go today with the AI Powered Equity ETF (AIEQ). It lets a machine look among standard financial measures, Twitter feeds, Reddit chat boards, and more and decide on its own mix of clues to favor. Performance for the four-year-old fund has been unremarkable, but Natividad believes the machine is getting smarter.
There is plenty of potential for AI mayhem, says Schmidt, who has advised the Defense Department on its use. The technology can keep watch for hypersonic missiles, which can arrive with little warning. “Imagine if it learns something wrong and it in fact makes the wrong recommendation and starts a war,” Schmidt says. He has called for China and Russia to swear off the use of automatically launched nuclear weapons.
Tools for making so-called deep fakes, or videos that can make prominent people appear to do or say anything, are open source or widely available. “The power of video is extraordinary,” Schmidt says. “If you produce a false video and you actually tell people it’s false, and you then show it to them, they still more or less believe it. That’s a human problem.”
And of course, AI doesn’t have to kill or destabilize to make us miserable. Schmidt says outrage is shared seven times more than reason. “Why are we surprised that we’re all upset about social media?” he says. He has written a book with Henry Kissinger, the former Secretary of State, and Daniel Huttenlocher, dean of the MIT Schwarzman College of Computing. It’s called The Age of A.I. and Our Human Future.
Schmidt compares AI with electricity and the telephone in its reach and capacity for both good and evil. AI will discover medicines that will save many millions of lives, he says.
“People always talk about, ‘Oh, the robots are going to take over the world,’ ” he says. “Not without us watching. And remember, we can always unplug them if we get really worried.”