The AI Bubble: Signs of a Pop and What Follows After

We are certainly in a bubble and we should be worried about the aftermath.

A few weeks ago, I was in Las Vegas speaking at an AI conference, surrounded by the usual buzz of excitement and innovation. During one of the evening networking events, I stepped up to the bar for a glass of wine. The bartender, a slender man named Ed in his mid-50s, greeted me with a grin that suggested he had a story to tell. As he poured my drink, he leaned in and said, "You know, I love this conference because I love AI. I bought NVIDIA stock three years ago, and after a few stock splits, I made $80k. I LOVE AI."

With that, he walked away with a sly smile, clearly proud of his investment and the fortunes it had brought him. His story perfectly encapsulates the fervor surrounding AI, but triggered a sense of anxiety for me. The story reminded me of that scene in the Big Short, when Mark Baum was talking to a boastful CDO manager over dinner and realizes he is in the midst of a subprime housing bubble. At that moment, I wondered how many times Ed had shared that story throughout the night. It struck me that we are indeed in an AI bubble.

How Do We Know We Are In a Bubble?

Ed the bartender is one tell-tale sign of a bubble - significant retail interest in a stock. However, there are a number of other signs I’ve personally noticed as a growth investor that we are in a market bubble:

• AI-Washing: companies rebranding to include ".ai" in their domain or claiming they use AI to ride the excitement wave. Some of them have succeeded in raising cheap capital.

• Skyrocketing Valuations: I covered semiconductors for a bit when I was an intern at a global investment bank in the Bay Area and the main take away about the semis industry is the cyclicality. We are definitely in the peak of a cycle just by looking at the EV/Sales multiples in the chart below. A number of fabless semis doubled or even tripled valuation multiples in the span of 1.5 years!

• Startups: you can also tell what the hottest sectors by looking at what YC is chasing. 50% of YC Winter '24 batch is building around AI. In the past, the cohorts were focused on the metaverse and crypto.

• Overly Optimistic Projections: the amount of value or ROI that needs to be generated by the CapEx spend is enormous and companies haven't been able to fill the gap. Sequoia estimates there is a return $125BN gap based on the money spent on chips. The gap is growing every quarter!

Select fabless semiconductor companies trading multiples on EV/Revenue (LTM) from CapIQ

What Happens When The Bubble Pops?

When the music stops, unfortunately, there will be a bloodbath. History tells us that bubbles rarely end without causing chaos (think dot-com crash of the 2000s and the housing crisis in 2008).

• Customer Exodus: One of the hallmarks of any bubble is the excitement surrounding grand visions and future possibilities. Customers have been promised revolutionary AI applications, even if they haven't demonstrated immediate commercial value. Customers of AI products will eventually realize that they aren't getting what they promised or aren't getting enough ROI. Users will slash budgets and speculative projects will be cut. Customers will focus on more practical, and immediate ROI applications.

• Industry Consolidation: retrenchment of customers will lead to winners and losers. Winners will be AI applications that have real product market fit, sustainable revenue models, and strong ROI for their customers. Companies that were founded on lofty ideas may become zombies, or acquired at bargain prices. The scale will tip in favor to well-capitalized firms with strong fundamentals and real customers. Those companies with cash, or a strong balance sheet will acquire or acquihire flailing companies.

• Funding Winter: generous VC funding is one of the reasons the AI sector is hot in the first place, leading to inflated valuations and ambitious growth targets. In the aftermath, investors will suddenly become more cautious and the scale will tip back to companies that are close to cash flow break even. There may even be a “flight to quality”, as investors focus on companies with established business models, and strong IP.

It is practically human nature to be excited by new technologies. I've personally lost on NFTs and crypto. However, one of the most immediate consequences will likely be a stock market correction, with the biggest losers being retail investors like Ed.

The Silver Lining

However, the fallout won't be all negative. I believe the AI industry will come out stronger at the end of the tunnel. I’m personally excited about the long term prospects that AI can bring to all businesses. I wrote about my enthusiasm before here. But the industry has recently spun out of control and a rationalization in AI will pave the way for a more mature, resilient industry. The survivors of the dot-com crash laid the foundation for growth in the next two decades. AI will likely follow a similar path.  The companies that survive the turbulence will be those with strong product market fit, and proven economic value. These players will be grounded in practicality rather than hype. As for when this will happen? Who knows.