Artificial intelligence has been a mega-wave breaking over global markets, with generative AI tools like ChatGPT capturing people’s imaginations more than previous applications. So far in 2023, AI has been one of the primary drivers of the stock market’s performance. But as investors wonder how to prepare for AI, most want to know, “What’s real and what’s hype?”
From an economic standpoint, we expect a meaningful increase in real GDP arising from higher productivity growth, particularly compared to today’s lackluster baseline. This should translate into faster earnings growth for the market overall, though the benefits will be distributed unevenly.
In fact, for most companies, the rewards from faster growth will be fairly limited. They’ll likely find themselves running to stay in place, spending much of their incremental revenue or cost savings on AI technology and staff who can harness it. While this may not happen immediately, it will unfold over time. In the meantime, their stocks may still rise in the nearer term, assuming investors extrapolate that growth past the point of sustainability.
As investors ask how to prepare for AI, we see two types of companies with the most to gain:
- Those for whom AI is an explicit driver of revenue and earnings growth, many of which have already experienced a surge in their stocks due to AI hype.
- Those who have a competitive position to integrate this technology into their business and can extract most of the gains for themselves while sharing some of the spoils with customers and partners.
Scaling the Economic Impact
To gauge AI’s impact, we first need to establish a baseline. The Congressional Budget Office estimates that over the next decade, potential GDP in the US will expand by approximately 1.8% per year.1 This breaks down to roughly 0.4% population growth and 1.4% productivity growth. Beyond the next decade, Bloomberg Economics forecasts that baseline growth in potential GDP will hover around 1.5%.
What could AI add to these figures? We believe that AI could credibly contribute an additional 1% per year, significantly increasing potential real GDP growth to 2.7% per year over the next decade and a half.
To arrive at that estimate, we’ve drawn from a recent McKinsey study that includes the effects of generative AI. Their research puts the potential productivity boost from AI and work automation between 0.6% and 3.6% per year from 2022–2040, depending on the pace of adoption. Generative AI alone accounts for 0.3 percentage points in the slowest implementation scenario and 0.7 percentage points in the fastest one.
The upper end of that range would outpace even the impressive productivity growth as the PC and internet took hold in the late 1990s and mid-2000s. If we assume realistic technology rollouts at around one-third of McKinsey’s fastest and slowest paces, that suggests roughly 1.6% of explicit automation and AI impact. However, since some of that is already implicitly baked into the baseline productivity estimate in GDP calculations (we estimate ~0.6%), we’ve only added 1% per year to avoid double-counting.
Using an extremely simple model to convert real GDP growth into S&P 500 earnings growth, an increase to 2.7% real GDP growth would boost S&P 500 earnings growth from 5.2% to 6.6%.
How High Does Your Boat Float?
While a rising tide tends to lift all boats, we expect some shareholders to benefit much more than others. For investors wondering how to prepare for AI, we see the rewards falling along two lines:
- Does AI explicitly create or expand the market for a given company’s products or services?
- Are they able to monetize tech boosts or must they simply run harder just to stay in place?
In many industries, we see AI simply raising the table stakes, effectively creating an arms race. Companies will invest in the technology not because they can outperform their competitors, but because if they don’t invest, their competitors will—with laggards left behind. Many companies are trying to whitewash that story in their conversations with investors, but as with past waves of technology and automation, they will likely follow a similar playbook.
Where AI’s Benefits May Stick
So how can investors prepare for AI? Let’s envision realistic scenarios where companies might harness AI, capturing the value it creates.
For instance, by using AI technology to streamline the drug discovery process, a biotech or pharma company could secure a critical patent granting them a formidable moat for years to come.
Shareholders of digital properties that can use generative AI for unstructured searches might also see a lift. For example, companies like Airbnb can leverage AI to make it easier for users to find a specific “vibe” in their diverse selection of rentals. Google and other search giants could tap generative AI to recommend itineraries that lead directly to bookings, potentially cutting out travel agents and aggregators. However, they would need to adjust their revenue model since they have historically charged a premium for top search placements. Additionally, a tech-savvy clothing retailer could use AI to filter unstructured searches and help customers find the exact shoes or jackets they want.
Healthcare providers may be able to deploy AI to improve the efficiency of cancer screening, radiology/imaging analysis (critical due to a shortage of radiologists), and surgeries. Doing so could substantially improve both public health and society. But where does the financial value accrue? First and foremost, to hardware makers integrating AI-driven software into their products—they already enjoy a natural advantage by owning reams of imaging data to train AI models. Further value will accrue to either the government, insurance companies, or medical providers, depending how reimbursements adapt to newfound efficiencies.
Agricultural equipment makers can also utilize AI. Incumbents like Deere and Agco have an edge when it comes to training autonomous vehicles given the potential trove of training data from their large installed bases. Yet even startups like Carbon Robotics can focus on specialized use cases like combining machine vision with lasers to remove weeds (check out this cool video).
Then there are the obvious AI names, many of which have already surged due to AI hype this year. Training generative AI models is incredibly compute-intensive and requires high-end chips for all that processing. If you’re a leading chipmaker or even running second in the market—like Nvidia and AMD—and hold a substantial technological edge, then a large profit pool awaits. Or perhaps, like Adobe, you make the AI-driven software-of-choice that creates significant efficiencies and benefits for your users. Combined with your vast, proprietary training data to prevent rights usage lawsuits, you’re also well positioned to surf the AI wave. Given that these two stocks have risen by over 200% and 50% year-to-date, respectively, the market seems fully aware of their strong fundamental prospects.
Where Could AI’s Impact Be Short-Lived?
Imagine a consumer goods company using ChatGPT or other generative AI to improve marketing and sales efficiency. While they’ll be able to save on labor costs or generate higher revenues for each dollar of labor, so too will their competitors. And unless competition is extremely disciplined, much of those gains will be sacrificed searching for other investment opportunities to maintain market share.
Now imagine a restaurant or retail store leveraging AI to improve inventory efficiency, which can boost sales and reduce the amount of capital tied up in slow-moving products. However, their competitors will also benefit from the same technology and processes. Zooming out, think of the industry overall. Will people buy substantially more burgers or burritos because of AI? Not likely. If the revenue pool stays roughly the same—and rivals benefit from similar technology and processes—your market share will hold steady, as will revenues. And while greater automation can lead to cost savings, the cost of the technology, including capital expenditures and depreciation, as well as any annual operating costs, will partially offset these savings.
As investors grapple with how to prepare for AI, they should note that competition works like erosion—a steady force whose impact becomes more apparent over time. In competitive industries (of which there are many), earnings may surge in the short run, causing returns on capital to exceed the cost of capital. This will incentivize competitors to spend more on marketing to open new locations or capture greater share. The level of increased spending or investment will depend on the competitive dynamic of a given industry. However, in many industries, this increased competition will drive returns on capital back toward the cost of capital, which can reduce margins and earnings in the medium term and offset any short-term benefits.
Positioning for AI
Overall, we expect the advent of AI to lead to faster economic growth, assuming it’s rolled out at a pace which is mildly (not massively) disruptive to the labor markets. It should boost overall efficiency and productivity, leading to higher average standards of living, and at least on its own, serve as a deflationary force.
Yet, two things remain uncertain: how these benefits will be distributed across the economy, and which workers will be positively or negatively affected. It will also be important to monitor how gains are split between labor and capital.
Ultimately, we expect faster real GDP growth to lead to higher earnings growth than in the absence of AI. While that should be a boon for the market overall, the spoils will be shared unequally. The companies that stand to benefit most are those that can either create a newly addressable market or harness the technology to boost revenues and profits. Investors have already bid up many of the former stocks. Other companies that could see a substantial lift will train AI on large sets of proprietary data while avoiding competition from equally well-positioned competitors. Of the remaining, many may be forced by rivals to utilize AI, only to see most of their rewards competed away.
Like others, we have our concerns about the potential consequences of AI, but we’re mainly excited by its potential progress. And as investment opportunities arise in the coming decade, our portfolio managers will be looking to seize them.
1 Potential GDP is defined as the amount of economic value the economy can produce without leading to higher or lower inflation.