Valued at 113x revenues. Really?
November 16, 2023
What we think about AI valuations hitting extraordinary highs
There is no doubt that AI is the most widely discussed topic of the year so far, both within the investment world and beyond. A report from McKinsey in June described AI as the “next productivity frontier” and applications like ChatGPT and GitHub Copilot have become popular extremely quickly. Generative AI in particular has seen a huge increase in investment over the past year or so. According to CB Insights, the number of Generative AI deals in Q3 2023 was slightly higher than the whole of 2022 at 170 and 168 respectively. What is truly remarkable, however, is that the value of those 170 deals was nearly six times higher than in 2022, increasing to $17.4bn from $3.2bn.
Revenue multiples for AI companies are extremely high. Perhaps the best-known AI company, OpenAI is currently valued at 22x its revenue, while Hugging Face, stability.ai and mosaic are valued at 113x, 100x and 65x, respectively.
The big question is how do you go about valuing AI companies and what makes them worth up to 113x their revenue?
A big factor in these incredibly high revenue multiples is the fact that all of the major tech companies are jostling for position in the industry. Of the ten largest deals done in the AI industry so far this year, a major tech company (think Microsoft, Amazon, NVIDIA, etc.) has been involved in all of them. When you have the largest companies in the world fighting to invest in the best AI companies, prima facie, it is no surprise that valuations are sky high.
The other reason, of course, is the enormous potential that AI will bring. Exactly when will vary for different use cases. There is no question that the signs are there though. I don’t know anyone who has not played with ChatGPT and with the time I’m already saving, I can see that with the right marketing, I will end up being willing to pay for the service.
In the AI market, companies often operate for a number of years before commercialising their IP and therefore might be loss making with very minimal revenues. Forecasts, in these circumstances, are often too speculative to be relied on (if they exist at all!) for say a tax valuation, but for a funding round, investors may see things differently. That is their right. Even if it is hard to evidence, we all know that the commercial potential is going to be huge and those who own the companies supplying AI solutions will make out like bandits.
In the meantime, there will also be M&A opportunities in AI which could bring investors stellar returns even if the company has not yet sold anything. Look at the valuation journey of Meta as an example.
Meanwhile with no extant revenues and large losses, multiples-based valuations are either impossible or highly speculative; any speculative valuation based on forecasts needs to be risk adjusted accordingly.
Balanced Scorecard valuation methodologies can be useful. They allow a valuation based on qualitative, not quantitative features. This creates a baseline valuation. You can then think about what else (the intangible value, macro-economic and “Investor Hope Premium” measures) must be applied on top to get to the adjusted valuation.
Forecasts are really valuable in understanding these other issues. We can look to see if the assumptions are reasonable and judge accordingly. There are always comparisons that can be found. Even if the industry is very new, we can look at other industries which may now be mature and see how companies in that sector were being valued during a similar stage in the cycle.
Earlier this year, we valued an AI start-up that had the potential to be worth billions, but at the time it was just a gang of bright computer scientists with a load of laptops trying to fix on exactly what the technology would be, let alone how it would be commercialised.
Our analysis showed that the current value of the company was a lot lower than the extraordinary valuation their investors were looking to invest at. We made sure to carefully examine and articulate the reasons that such a high Hope Premium could be applied to the business.
Will AI valuations suddenly crash? Probably not yet. There is too much interest and some AI businesses are beginning to commercialise. However, when any industry reaches these levels of excitement some will win, and others will lose. There is a good reason why Nvidia as a chip manufacturer is storming ahead – in a gold rush the people who sell picks and shovels always win.
So for AI companies, funding round valuations at over 100x (predicted) revenues will endure for a while yet. It’s worth bearing in mind that if a tax valuation is required, it will probably be necessary to provide evidence and arguments as to why these pre and post money valuations should be disregarded and why a different approach is more appropriate.
