Over the four decades of radical change that began with the PC-driven shift of the 1980s, the annual return on stocks has been only marginally higher than in the five preceding decades. AI will have a similar impact: it will produce a few big winners, but its net positive effect on markets overall will be surprisingly small.
NEW YORK – Over the past year and a half, artificial intelligence has gone from imaginative speculation to market reality, fueling both greed and fear. As a few companies, such as Nvidia and Microsoft (with its $10 billion investment in OpenAI), gained hundreds of billions of dollars in market capitalization, Hollywood writers and actors went on strike, demanding new rules for AI-generated material.
I am both a believer and a skeptic of revolutionary changes in markets, having seen major disruptions play out both in my personal life and in my investment portfolio. After personal computers in the 1980s came the dot-com boom and the internet revolution in the 1990s, followed by smartphones in the 2000s, which facilitated the explosion of social media in the 2010s. The defining feature of these developments was not just that they affected broad swaths of business – sometimes positively, sometimes adversely. It was that they fundamentally changed how we live, work, and interact.
AI holds the potential to drive similarly sweeping changes. My wife, who teaches fifth grade, is already grappling with students using OpenAI’s ChatGPT to complete their homework assignments, and companies are scrambling to integrate AI into their operations. For investors, however, the task is to separate hype from reality. We can start by revisiting the big technological and market developments of the last four decades.
The first of these was the PC-driven shift of the 1980s, which coincided with a decade of high stock-market returns, as did the 1990s dot-com boom. Stocks then flatlined in the aughts of this century – making it one of the worst decades in the market’s history – before performing well again in the 2010s, when Big Tech was the big winner. Yet over these four decades of radical change (1980-2022), the annual return on stocks has been only marginally higher than in the five decades prior.
I believe that AI will have a similar impact: its net positive effect on markets overall will be surprisingly small. The reason is that there will be only a few big winners, but many wannabes, losers, and market chasers who are forced to invest in the new technology merely to keep up. Such was the case with each of the revolutionary shifts of the last four decades. We have seen the same paradigm play out in software, online commerce, smartphones, and social media.
Notably, the early leaders in these businesses have often fallen by the wayside. In the current hype cycle, almost every company is trying to don the AI mantle, just as every company in the 1990s aspired to be a dot-com, and just as many companies in the last decade claimed to be “user-intensive” platforms. From an investment perspective, separating the wheat from the chaff will become only more difficult in the coming years; but that is part of the learning process.
Access every new PS commentary, our entire On Point suite of subscriber-exclusive content – including Longer Reads, Insider Interviews, Big Picture/Big Question, and Say More – and the full PS archive.
Subscribe Now
The market is littered with the carcasses of formerly successful businesses that were disrupted by technological change. Investors in these companies not only lose money as the displacement occurs; they often invest even more because they are attracted by the asset’s apparent “cheapness.” To cite just two examples, this happened to investors in the brick-and-mortar retail companies that were devastated by online retail, and to investors in the newspaper and traditional advertising companies that were upended by online advertising.
If AI meets its disruptive potential, many businesses will be similarly upended – not least consulting and banking. If you are in the business of offering generally formulaic, mechanical advice, you are vulnerable to replacement by a machine, though the timeframe for such changes remains unclear. Thus, even if we accept the premise that AI will radically change the way businesses and individuals behave, there is no low-risk path for investors to monetize this belief.
But should we even accept that premise? In the hype phase, AI is oversold as the solution to every problem, real or imagined. It is used to justify large price premiums for companies in its orbit, usually without any attempt to quantify or back up such valuations. Those selling these premiums would argue that there is too much uncertainty about AI’s potential to offer concrete numbers. But if you are paying a high price for a company’s “AI effect,” you would do well to try. With sound judgment (and data), you can still map out scenarios for how AI might affect future cashflows, growth, risk, and ultimately value.
The AI Ecosystem
In making such estimates, it helps to consider which part of the AI ecosystem a company inhabits. One important element is hardware and infrastructure. Every major change over the last few decades has brought new hardware and infrastructure requirements, and AI is no exception. Nvidia owes its AI effect to the increased demand for AI-optimized computer chips. Since that market is expected to grow rapidly, Nvidia will have higher revenues and earnings.
Then there is software. AI hardware, by itself, has little value unless it is paired with code that can take advantage of the computing power it offers. AI software can take many forms, from bespoke AI platforms and chatbots to the underlying foundation models (such as the deep-learning algorithms that power image and voice recognition and natural language processing). While there is less structure and more uncertainty in this area, it potentially has a much greater upside than hardware.
Another part of the ecosystem is data, which AI requires on an immense scale. There will be some businesses that create greater value from collecting and processing data specifically for AI applications. Big data, which was used more as a buzzword than a business proposition over the last decade, may finally find its place in the value chain. But this path will be nonlinear and unpredictable.
A final area is the sprawling field of AI applications, which is relevant to all the companies that will be consumers of AI rather than purveyors. The sales pitch from AI promoters is that it will reduce costs (primarily by replacing human labor) and increase efficiency, thus boosting profitability. But even if one concedes the first claim (even though AI automation will be neither as efficient nor as cost-reducing as promised), one should be quite skeptical of the second. After all, if every company uses AI to reduce costs and increase efficiency, none will be better off on a relative basis. The most likely outcome is lower prices for their products and services, not higher profits.
Thus, if AI delivers as promised, it will actually make companies less profitable in the aggregate. Consider active investing. If your job is to deliver excess returns, you may briefly gain an edge by deploying AI to spot patterns in the data that are indiscernible to an individual human. But soon enough, the same tools will be available to every investor, and your edge will be gone.
A final question concerns AI’s social effects. Will it make our lives easier or more difficult? Will it make the world better – for example, fairer and more sustainable – overall, or worse? Some AI advocates promise a utopia where algorithms and robots will eliminate drudgery and toil and bring an unbiased eye to data analysis. Others argue that AI will be an all-powerful tool with which big companies and governments will control minds and acquire greater power. I believe both are right. AI will be a positive addition to some occupations and aspects of our lives, and it will also create unforeseen adverse consequences.
As for the idea that AI can be held in check and made to serve only noble motives through policy restrictions or regulation, I am not optimistic. Any law aimed at curtailing AI’s excesses will almost certainly set in motion unintended consequences, some of which could prove even worse than the problem that was supposed to be solved. Having seen how badly regulators and legislators handled the consequences of the social-media explosion, I doubt they would even know where to start with AI.
While my perspective is admittedly pessimistic, I believe it is realistic. As with social media, it will be up to us as consumers to try to draw red lines and separate the good from the bad. We may not succeed, but no one else will do it for us.
To have unlimited access to our content including in-depth commentaries, book reviews, exclusive interviews, PS OnPoint and PS The Big Picture, please subscribe
NEW YORK – Over the past year and a half, artificial intelligence has gone from imaginative speculation to market reality, fueling both greed and fear. As a few companies, such as Nvidia and Microsoft (with its $10 billion investment in OpenAI), gained hundreds of billions of dollars in market capitalization, Hollywood writers and actors went on strike, demanding new rules for AI-generated material.
I am both a believer and a skeptic of revolutionary changes in markets, having seen major disruptions play out both in my personal life and in my investment portfolio. After personal computers in the 1980s came the dot-com boom and the internet revolution in the 1990s, followed by smartphones in the 2000s, which facilitated the explosion of social media in the 2010s. The defining feature of these developments was not just that they affected broad swaths of business – sometimes positively, sometimes adversely. It was that they fundamentally changed how we live, work, and interact.
AI holds the potential to drive similarly sweeping changes. My wife, who teaches fifth grade, is already grappling with students using OpenAI’s ChatGPT to complete their homework assignments, and companies are scrambling to integrate AI into their operations. For investors, however, the task is to separate hype from reality. We can start by revisiting the big technological and market developments of the last four decades.
The first of these was the PC-driven shift of the 1980s, which coincided with a decade of high stock-market returns, as did the 1990s dot-com boom. Stocks then flatlined in the aughts of this century – making it one of the worst decades in the market’s history – before performing well again in the 2010s, when Big Tech was the big winner. Yet over these four decades of radical change (1980-2022), the annual return on stocks has been only marginally higher than in the five decades prior.
I believe that AI will have a similar impact: its net positive effect on markets overall will be surprisingly small. The reason is that there will be only a few big winners, but many wannabes, losers, and market chasers who are forced to invest in the new technology merely to keep up. Such was the case with each of the revolutionary shifts of the last four decades. We have seen the same paradigm play out in software, online commerce, smartphones, and social media.
Notably, the early leaders in these businesses have often fallen by the wayside. In the current hype cycle, almost every company is trying to don the AI mantle, just as every company in the 1990s aspired to be a dot-com, and just as many companies in the last decade claimed to be “user-intensive” platforms. From an investment perspective, separating the wheat from the chaff will become only more difficult in the coming years; but that is part of the learning process.
Subscribe to PS Digital
Access every new PS commentary, our entire On Point suite of subscriber-exclusive content – including Longer Reads, Insider Interviews, Big Picture/Big Question, and Say More – and the full PS archive.
Subscribe Now
The market is littered with the carcasses of formerly successful businesses that were disrupted by technological change. Investors in these companies not only lose money as the displacement occurs; they often invest even more because they are attracted by the asset’s apparent “cheapness.” To cite just two examples, this happened to investors in the brick-and-mortar retail companies that were devastated by online retail, and to investors in the newspaper and traditional advertising companies that were upended by online advertising.
If AI meets its disruptive potential, many businesses will be similarly upended – not least consulting and banking. If you are in the business of offering generally formulaic, mechanical advice, you are vulnerable to replacement by a machine, though the timeframe for such changes remains unclear. Thus, even if we accept the premise that AI will radically change the way businesses and individuals behave, there is no low-risk path for investors to monetize this belief.
But should we even accept that premise? In the hype phase, AI is oversold as the solution to every problem, real or imagined. It is used to justify large price premiums for companies in its orbit, usually without any attempt to quantify or back up such valuations. Those selling these premiums would argue that there is too much uncertainty about AI’s potential to offer concrete numbers. But if you are paying a high price for a company’s “AI effect,” you would do well to try. With sound judgment (and data), you can still map out scenarios for how AI might affect future cashflows, growth, risk, and ultimately value.
The AI Ecosystem
In making such estimates, it helps to consider which part of the AI ecosystem a company inhabits. One important element is hardware and infrastructure. Every major change over the last few decades has brought new hardware and infrastructure requirements, and AI is no exception. Nvidia owes its AI effect to the increased demand for AI-optimized computer chips. Since that market is expected to grow rapidly, Nvidia will have higher revenues and earnings.
Then there is software. AI hardware, by itself, has little value unless it is paired with code that can take advantage of the computing power it offers. AI software can take many forms, from bespoke AI platforms and chatbots to the underlying foundation models (such as the deep-learning algorithms that power image and voice recognition and natural language processing). While there is less structure and more uncertainty in this area, it potentially has a much greater upside than hardware.
Another part of the ecosystem is data, which AI requires on an immense scale. There will be some businesses that create greater value from collecting and processing data specifically for AI applications. Big data, which was used more as a buzzword than a business proposition over the last decade, may finally find its place in the value chain. But this path will be nonlinear and unpredictable.
A final area is the sprawling field of AI applications, which is relevant to all the companies that will be consumers of AI rather than purveyors. The sales pitch from AI promoters is that it will reduce costs (primarily by replacing human labor) and increase efficiency, thus boosting profitability. But even if one concedes the first claim (even though AI automation will be neither as efficient nor as cost-reducing as promised), one should be quite skeptical of the second. After all, if every company uses AI to reduce costs and increase efficiency, none will be better off on a relative basis. The most likely outcome is lower prices for their products and services, not higher profits.
Thus, if AI delivers as promised, it will actually make companies less profitable in the aggregate. Consider active investing. If your job is to deliver excess returns, you may briefly gain an edge by deploying AI to spot patterns in the data that are indiscernible to an individual human. But soon enough, the same tools will be available to every investor, and your edge will be gone.
A final question concerns AI’s social effects. Will it make our lives easier or more difficult? Will it make the world better – for example, fairer and more sustainable – overall, or worse? Some AI advocates promise a utopia where algorithms and robots will eliminate drudgery and toil and bring an unbiased eye to data analysis. Others argue that AI will be an all-powerful tool with which big companies and governments will control minds and acquire greater power. I believe both are right. AI will be a positive addition to some occupations and aspects of our lives, and it will also create unforeseen adverse consequences.
As for the idea that AI can be held in check and made to serve only noble motives through policy restrictions or regulation, I am not optimistic. Any law aimed at curtailing AI’s excesses will almost certainly set in motion unintended consequences, some of which could prove even worse than the problem that was supposed to be solved. Having seen how badly regulators and legislators handled the consequences of the social-media explosion, I doubt they would even know where to start with AI.
While my perspective is admittedly pessimistic, I believe it is realistic. As with social media, it will be up to us as consumers to try to draw red lines and separate the good from the bad. We may not succeed, but no one else will do it for us.