Big fund-raising rounds and high valuations have some wondering whether the AI sector is in a bubble in the nature of the dotcom boom. As of this writing, OpenAI is valued at over $80 billion; Amazon added another $2.75 billion to its investment in Anthropic; and even some very early-stage startups, like France-based Mistral AI, have racked up hundreds of millions in venture-capital funding at valuations over a billion dollars. Alphabet CEO Sundar Pichar has said AI could be more profound than the invention of electricity or the discovery of fire. Is the hype real?
It’s too early to tell and we’re not market prognosticators. But here are few words of caution. There is a huge range of business models in the AI space. Some are variations on old themes: consumer-facing tools and toys (like chatbots and search engines) that make money from subscriptions (often baited with freemium versions) and advertising dollars. Others are more far-reaching—and more untested—such as enterprise tools that promise to change the way we do business and organize our society. It’s quite possible that some fairly vanilla business models, with limited upside, are being overvalued because they benefit from the glow of AI.
Another fundamental risk has to do with the potentially winner-take-all nature of AI. A general-intelligence tool, if and when one is developed, might be able to do basically everything that all its specific-intelligence competitors do. This could lead to a profound consolidation of the market, which would counsel in favor of getting behind the current market leaders or businesses that are well positioned to be acquired by or to serve those leaders.
This consolidation could well be spurred, more immediately, by forthcoming regulation (and, inevitably, enforcement), which will raise the cost of doing business, likely more, on a relative basis, for smaller players than big ones. Similarly, consolidation could also be prompted by IP law. It is an open question whether the data that is used to train generative AI is subject to fair use or must, instead, be licensed, as copyright owners such as the New York Times contend in lawsuits challenging the use of their IP. If large-scale AI training on copyrighted material turns out to be fair use, that will be good for startups and others that don’t necessarily have the resources to negotiate deals with content owners. If, on the other hand, licenses are needed, that could mean more consolidation, to the benefit those who hold the licenses.
AI is perceived by some as having potentially human-species-wide ramifications. OpenAI, the most valuable AI business, has been challenged by Elon Musk for having strayed from its not-for-profit mission to safely create AI that would benefit humanity. We are not aware of any industry where the market leader is committed, in its very charter, to such ideals. If the sentiment that AI should benefit humankind catches on among the public or lawmakers, or if AI is perceived as too dangerous, that could lead to a major reordering of the industry.
Until then, however, the market is hot and the question is whether the hype is real. To answer that immediate question companies should be evaluated on their fundamentals: product-market fit, past performance, the size of new opportunities, the abilities of management, and the defensibility of their technology. Just as Amazon emerged from the dotcom boom and bust, huge valuations will be won, and lost, over the course of the ongoing AI boom. The dotcom boom ended in bust for many, but for a select few, long-term booming businesses were born.
[View source.]
