Artificial Intelligence: What Investors Should Know Now
The last 125 years have seen the emergence and explosive growth of many different technology-driven industries, including automobiles, airlines, filmed and broadcast communications and entertainment, pharma, biotech, logistics, and, of course, personal computers and the internet. Some, notably the airlines, have struggled to create shareholder value, while others have created trillions in wealth. We are witnessing the birth of artificial intelligence as a practical reality with tremendous potential. But how should an investor think about AI, and what do they need to know about AI’s risks and opportunities? This edition of the Chronicle will try to tackle these critical questions.
The Past is Prologue—Sometimes
A chart of Cisco Systems (CSCO) during the Internet boom of the late 90s is a cautionary tale. Cisco was the preeminent maker of internet routers and networking systems. In the 18-month period between September 1998 and March 2000, Cisco’s share price rocketed from about $11 per share to over $47. One year later, it was back at $11. Cisco was the darling of the “irrational exuberance” period, when internet stocks went crazy.
Cisco was not the only casualty of the “tech wreck” bear market of 2000 – 2002. The NASDAQ index, which tracks the value of tech stocks, rose fivefold between 1995 and 2000. However, the bubble burst and the index fell 77% by October 2002. Unlike many others, Cisco survived and today has a market cap above $240 billion, but it’s been far eclipsed by other tech players. The internet craze was a wild time on Wall Street, but from our vantage point in 2025, can there be any doubt that the internet has had a profound impact and created trillions of dollars of wealth for investors?
While we can’t know anything with certainty, it seems pretty clear that AI is going to be a massive driver of wealth. AI holds tremendous promise in healthcare, manufacturing, supply chain management, education, e-commerce, defense, transportation, and defense—to name a few industrial sectors.
Share Prices Are High, but Are They Irrational?
It is clear that AI holds great promise. Determining what portion of the Information Technology (IT) sector’s market capitalization is attributable to AI, however, is tough. This is a complex equation, as many companies integrate AI into their operations and revenue models. We suspect that over the long haul, AI will simply be in everything and every company, in much the same way the internet is now. Even a coal-fired utility is dependent upon the internet for virtually every aspect of its business, from electricity generation to payroll. Most likely, AI will be like that.
The top five AI companies as of now are Apple, NVIDIA, Microsoft, Alphabet (Google), and Amazon. Collectively, these five have a combined market capitalization of almost $15 trillion, nearly 29% of the entire S&P 500 index. The other 62 companies that comprise the IT sector within the index have a combined market cap of about $400 billion. With 97% of the sector’s market cap, the big five are clearly the tail that wags the dog.
As of February 3, 2025, the price-to-earnings (P/E) ratio for the Information Technology sector within the S&P 500 was approximately 36.65. This figure is notably higher than the sector’s 5-year average P/E range of 25.41 to 32.29, indicating that the sector is richly valued. Price is only one way to think about these things; consider also that over the past decade, the IT sector delivered an annualized return of 19.2%. The broader market returned 12.3% in the period.
This sector is not only expensive, it’s volatile. The ten-year volatility has averaged 18.3%, while the overall S&P 500’s figure was 14.9%. On the bright side, consider that 22% more volatility produced 56% more return. Before you fully succumb to FOMO, consider that on January 27, NVIDIA, arguably the pure play in AI, fell 17%, erasing $590 billion of market cap, the all-time record one-day loss in U.S. stock market history.
How big a loss was that? Fifty-five of the sixty-seven IT sector companies have market caps below $590 billion. Now think about this statistic: after the one-day bloodletting, NVIDIA still had a P/E north of 47. Investors obviously have very high expectations for AI.
While optimism is high for AI, we are nowhere near the irrational exuberance level of the late 90s. At the peak of the dot-com bubble in 2000, the median P/E ratio of the top 10 U.S. companies by market capitalization was around 60x. Apple’s P/E on February 4, 2025 was 37.87 and Microsoft’s was 33.28. We are a long way from a bubble, historically speaking.
Big Impact Does Not Always Equal Big Profits
Based on everything we know, AI has the potential to be highly profitable for investors, whereas industries like airlines have historically struggled to maintain long-term profitability. There are a number of reasons for optimism about AI.
High Margins and Scalability
Software companies generally have high gross margins, and AI, in particular, could be a cash cow. Once an AI model is trained, it can be deployed across millions of users with near-zero additional costs. Contrast that with airlines, which must spend big money on aircraft, personnel and fuel on every flight.
In 2007, Warren Buffett wrote in his annual shareholder letter, “If a farsighted capitalist had been present at Kitty Hawk, he would have done his successors a huge favor by shooting Orville down. Investors have poured money into a bottomless pit, attracted by growth when they should have been repelled by it.”
Airlines and other transportation industries operate on low margins due to high fuel costs, labor expenses, maintenance, and regulatory fees. They also rely on physical assets that are expensive to maintain. By contrast, AI has virtually none of those costs. After initial development, it can scale to the moon.
Low Capital Expenditure vs. High Fixed Costs
Most AI companies are asset-light—they invest heavily in research and computing power but don’t need to maintain factories, rolling stock, or supply lines. They don’t require the massive investment in plant and equipment like the automobile and energy industries do. That’s the thing about fixed costs: they hit you wherever you are in the demand cycle.
Recurring Revenue Model. Many AI-powered businesses will operate on a subscription or licensing model—think software as a service and cloud services—providing consistent and predictable high-margin revenue. By contrast, many other industries’ revenues are transactional and cyclical. AI revenues should be more reliable and less subject to seasonal and cyclical variations in demand.
Lower Price Competition & Stronger Differentiation
AI businesses can create strong competitive advantages through proprietary algorithms, unique datasets, and network effects. A great example of network effects is the telephone, which becomes increasingly valuable, even to existing users, as more people join the network.
We can already see, in these early days, that network effects benefit companies like OpenAI and Google’s Gemini. AI platforms, like ChatGPT, and megadata applications, like self-driving cars, improve over time as they collect more data. This is the capitalist’s dream, competing with relentlessly improving technology and capabilities, not on price.
Exponential Growth Far Into the Future
It’s reasonable to expect AI to follow a Moore’s Law-type trajectory, where computing power doubles every 18-24 months. Although AI’s growth may not continue its exponential growth the way chips did from 1965 to 2025, AI has seen tremendous growth since 2018.
Moore’s Law forecasts hardware improvements (transistors per chip), but AI benefits from both new chip technology and algorithmic efficiency gains. Companies like NVIDIA and AMD are building AI-specific processors that accelerate model development, while new AI architectures, like transformers and mixture of experts, demand fewer computations per training cycle even as their accuracy improves.
AI models today achieve better performance using less training data and lower computational costs, much like how Moore’s Law enabled more efficient computing. AI is a very technical field, but this is a big advantage. Exponential growth often means rapid profit scaling, which could justify today’s high multiples.
Strong Geopolitical & Economic Tailwinds
Governments and businesses are pouring billions into AI development due to its role in national security, automation, and economic growth. In its first week in office, the new administration announced StarGate, a $500M private investment in AI data centers, a huge gain in infrastructure.
Not a day goes by when AI isn’t in the news. As a story, and now even a meme, AI has it all: economics, politics, culture, foreign policy, energy, healthcare, climate change, and potentially breathtaking wealth creation. Precedence Research forecasts that AI’s market size will grow from $638.23 billion in 2024 to approximately $3.68 trillion by 2034, and McKinsey & Company suggests that the total addressable market for AI could reach $15 trillion, considering the broad economic impact of AI across various sectors. Like the internet before it, AI is poised to change everything.
No Free Lunch
The great promise of AI, both for humanity and investors, is not without its potential risks. Automobiles, airlines, and computers were all greeted with a frenzy of entrepreneurial activity. There have been at least 1,900 automobile companies, producing over 3,000 different makes of American cars, since the Duryea in 1895. People originally thought that pet food had the big potential of the internet, not seeing then what social networking would become.
These are early days, and we simply cannot predict who the winners and losers of AI will be. However, it seems a safe bet that there will be both winners and losers as the industry matures and the marketplace votes with its dollars.
As the automobile essentially destroyed the horse as an industry, AI is sure to have an impact on jobs. We are already seeing AI disrupt customer service, data entry, marketing, and even software development. On the other hand, AI could create entirely new job classifications in machine learning and prompt engineering. These are early days, but change is coming.
Governments are sure to have concerns about bias, misinformation, job displacement, and security. The Biden administration’s recent aborted attempt to corral AI is just one example of what could happen. As it regulated airlines, energy, transportation, and the airwaves before it, the government seems likely to claim some regulatory role in AI.
As mentioned above, the internet is a part of everything now. An investment of $100 in the S&P 500 at the beginning of 1998, with dividends reinvested, would have grown to approximately $1,011.11 by the end of 2024. Who can say how much of those gains was due to the internet? Whatever its ultimate impact, it seems clear that AI has the potential to power a more productive world, and that will likely be a good thing for investors holding diversified portfolios.
