ChatGPT has been all the talk, not just on the internet but in corporate meeting rooms and among investment bankers, hedge fund analysts, and other financiers struggling to determine what this all means for equity and debt valuations (there are also questions about artificial intelligence’s impact on finance itself, but that’s a separate, though important, topic). Most in the crosshairs are Alphabet (GOOG, GOOGL) and Meta (META), although all tech firms are closely exposed to the risk and opportunity of AI.
GOOG learned this the hard way last week, when its shares dropped by $100 billion in market cap after its chatbot made a small factual mistake. Since the error was about a fact, and not a hard one to look up (what telescope was the first to take a photo beyond the solar system?), it looked like a fatal blow to Alphabet’s wheelhouse: the discovery and delivery of relevant and accurate information.
ChatGPT has made similar errors, and arguably the hit in valuation is partly due to poor quality control from Alphabet management than from the mistake itself. Not catching such an embarrassing mistake will put any management into question, which demonstrates the complexity of valuing our AI future into equities. How much of the mistake is fixable, how much of that fixable-ness was priced into the drop, and how much of that is separate from a hit in valuation due to management’s big mistake?
Looking into the future, more and similarly complicated questions arise. Will Alphabet get its AI fixed in time to compete with Microsoft’s (MSFT) soon to be unveiled AI-driven search engine? How will the new race to get the best AI chatbot/search engine impact these companies’ future revenue and earnings? What about competitors? And how much is an AI-driven search engine or chatbot really worth?
These questions cannot be answered with the conventional tools of finance—at least not on their own. Financial calculations rely on a historical precedent that can be used to project into the future; there is no historical precedent for AI, and comparables like the emergence of the internet or the PC are poor metaphors.
Financiers are busy struggling to understand this AI, how it functions, what it costs, and what it possibly can do. That last question is particularly challenging, because the answers can get extremely fanciful (“AI will start a nuclear holocaust like in the Terminator movies”) or extremely prosaic (“AI can improve search engine speed and accuracy by X% in Y years”). The latter kind of answer will look more mature, and writing about such details in a report to clients won’t get any analyst fired. However, limiting oneself to just that end will inevitably lead to a low valuation bias that could theoretically misvalue equities by trillions of dollars.
Of course, going to the other extreme will result in pablum with little intellectual or economic value. The analyst must follow her instincts to find a middle path between these two extremes, which ideally can be found with the insights of proper due diligence. An analyst who wants to properly value AI will begin talking to AI experts, reading their analyses of AI, and incorporating that knowledge into a new model projecting the economic value, run rate, and total addressable market of this new technology and its use cases. This is harder than it sounds; the analyst will need to know which experts are legitimate and which are not, which is hard to do in any field but particularly challenging in technology.
When good due diligence is met with solid financial acumen, analysts can predict the future—and make a lot of money doing so. And the rewards of predicting the future when it comes to brand new technologies, especially ones both very hyped up and seeing widespread adoption, can be significant.
