Investors in public and private markets are indiscriminately offering huge premiums to AI companies, and the party continues – the music is still playing and people are still dancing. But when the music stops, investors scramble for assets in a frenzied game of musical chairs.
We’ve seen this all before. Similar to the dot-com bubble, the long-term potential of this technology is enormous. The hype is just part of the initial hype. We are already in a valley of disillusionment, and once the AI bubble bursts, the flaws in our valuations will be exposed.
The unhealthy entanglement of AI companies
Some of the largest AI companies have a pattern of suppliers investing in their customers, creating a widespread issue of circular investing. For example, Nvidia is investing in OpenAI, so Nvidia directly or indirectly supplies GPUs to OpenAI. [graphics processing units] And this investment will allow OpenAI to buy even more products. A similar pattern can be seen with Nvidia’s investment in Coreweave. Coreweave buys GPUs from Nvidia and sells GPU capacity to Open AI and other Large Language Model (LLM) providers.
Meanwhile, major cloud providers have invested in OpenAI and Anthropic to sell cloud computing for training and inference. This pattern is repeated downstream, allowing OpenAI and other LLM providers to invest in other companies that can build applications on top of OpenAI’s ChatGPT.
Such intertwined partnerships are risky and unhealthy for the entire ecosystem. While the biggest companies can absorb the drop in valuations, other AI companies are being dragged down by the hype, especially in private markets.
For example, Lovable’s $330 million Series B was valued at $6.6 billion, and Mistral AI’s €1.7 billion Series C was valued at €11.7 billion, so the bar is already set high. There is nothing wrong with valuing individual companies in this way, but it creates unrealistic expectations for investors and entrepreneurs. To generate a 10x return on a Series B investment valued at $6.6 billion, the company would need to be sold for $66 billion. To put this into context, there are fewer than 10 public cloud software companies with market capitalizations above $60 billion.
While there is no doubt that GenAI will be widely adopted and create the next generation of businesses that can achieve such valuations, it is problematic for investors to value all startups as breakthroughs when 99% of companies do not necessarily.
Annual recurring revenue (ARR), but not what we know
Another flaw in AI company valuations is ARR reporting. This is the biggest driver of value for software companies, but it’s not what it used to be. Previously, it was based on subscription accruals, but now includes many other measures that are much more difficult to predict, such as one-time contracts, volume-based, performance-based, and value-based contracts.
Beyond the contract structure, there are other qualitative aspects to keep in mind. Short sales cycles and short implementation times are benefiting the growth of AI companies. However, this rapid sales could be due to hype and excitement, and a market targeting individuals (not corporate customers) who are less likely to renew. Also, a short implementation time may indicate that the technology is not significantly increasing productivity.
Gross margins are under pressure beyond repair.
Many AI companies operate at high costs and have overly optimistic revenue projections. Because AI models are expensive to develop, train, and maintain, many operate at very low or even negative profit margins. Traditional software companies typically have gross profit margins in the 70% to 80% range. On the other hand, many AI companies’ true gross profit margins are in the low teens to 20s.
One solution is to grow out of this situation quickly and hope that the costs of the model come down fast enough that the profit margins are low. However, so far this has not materialized and the high costs have not come down.
Bottom line: focus on the fundamentals, not the music.
As AI companies’ valuations soar, we need to focus on fundamentals. Current expectations are too high and the party is spiraling out of control. Apart from the irrational exuberance, there are still smart investments to make. There is a significant investment opportunity for B2B software companies looking to replace existing enterprise solutions based on AI-native capabilities.
There is an even greater opportunity for agent AI software companies to automate large portions of the existing professional services market, which is at least 10 times the size of the current cloud software market. These companies will impact productivity and transform business processes that are currently dominated by manual processes. These solutions require buy-in from multiple parties, which means the sales process becomes more difficult. But in the long run, you’ll get more consistent and recurring revenue.
As with the cloud era, most of these companies won’t become $100 billion or even $10 billion businesses, but that’s fine as long as the investment is based on realistic expectations and reasonable entry valuations. There is an unprecedented opportunity here to create long-term sustainable value for investors.
The opinions expressed in Fortune.com commentary articles are solely those of the author and do not necessarily reflect the author’s opinions or beliefs. luck.
