Nvidia revenue outlook confirms AI bonanza

Is Nvidia the epic outlier that only comes around once every decade?

Nvidia said that the underlying demand for their AI chips has just started with the CEO Jensen saying this was the beginning of a 10-year cycle. While these comments of course are genuine they also reflect a poor understanding of the potential bubble AI related stocks such as Nvidia is part of. A comment like this will just add to the bubble dynamics sucking many retail investors into the stock which is already reflecting huge expectations. The issue that nobody has any idea of where we are in 10 years from now. The outcome range for AI chips is extreme and because of this significant variance the pricing of Nvidia is almost impossible leading to a high probability of mispricing and even a bubble-like dynamics.

The enterprise value of Nvidia is up as much as the upward revision to the 12-month forward revenue meaning there is no change in the forward valuation of Nvidia post the earnings release. The 12-month forward EV/Sales ratio is still 22.3x compared to 1.1x around 10 years ago reflecting both the strong underlying growth dynamics and improved operating margins, but even more importantly investors’ extrapolation of AI technology into the future as something that is more or less unbounded. Nvidia has the highest 12-month forward EV/Sales of any stock in the S&P 500 Index.

The 12-month forward EV/Sales ratio is one of the best quant factors in backtesting with a significant Q-ratio, meaning that the return spread between the lowest and highest decile groups of stocks is quite significant. In other words, stocks with the highest EV/Sales ratio tend to significantly underperform stocks with the lowest EV/Sales ratio. These observations are naturally a phenomenon of aggregation and there will always be outliers. Amazon was one of those outliers for almost two decades. So maybe Nvidia is also one of these occasional outliers defying the aggregate observations. The important difference between Amazon and Nvidia is that Amazon defied the aggregate dynamic because it came from a small size swimming into a vast ocean of growth, while Nvidia is starting in a pole position on valuation with a market value already close to $1trn.

The market opportunity and the risks lying ahead for AI

In Nvidia’s investor presentation from this February the chipmaker says that the market opportunity is around $1trn in revenue. Whether this figure is the total revenue in these categories or the revenue opportunity available for Nvidia with some x market share is not clear. But the slide helps shape a particular narrative about how big they see this market. For reference the global economy is expected to grow to $106trn in current USD prices this year.

However this market opportunity does not come without risks and it is our obligation as investment analysts to point out these risks. In our equity note Great opportunities and risks lie ahead for semiconductors we highlight two key risks, but we should add another one here, so the three key risks for Nvidia and the overall AI and semiconductor theme are these:

1. The US and China tension driving US export restrictions on semiconductors to China could escalate further severely damaging the global supply chain of semiconductors. This very risk was highlighted by Nvidia CEO Jensen in a big FT interview yesterday before the earnings announcement.
   
   
2. PFAS chemicals are very important for the manufacturing of wide range of consumer applications including semiconductors, but these substances have been found to cause cancer and other severe illnesses among humans. 3M is ending production of PFAS by the end of 2025 and the EU is introducing a ban with a 13-year transition period. But the industry has no cheap and obvious alternative to PFAS so production of microchips may become more expensive in the future.
   
   
3. Electricity consumption is going to rapidly increase in the years ahead driven by electric vehicles, heat pumps replacing gas boilers for heating, and data centers. If an explosion in AI adoption takes place then the need for electricity will grow further and there is a real risk that both electricity production and grid upgrades and expansions will not be able to keep up creates a physical barrier for the semiconductor and AI industry.