Bubble Duration

Bubbles serve an important purpose but they eventually resolve.

I consider the 1995 to 2000 bubble to be my magnificent moment. What a time to be a technology geek and a stock analysis junkie. It was a glorious time that continued to surprise people with its power and duration. Even Stan Druckenmiller got crushed by it.

My bubble started with some pure technology-related drivers even before the internet arrived. New forms of computing (PCs, Minis, networks, and distributed architecture) were fueling a significant increase in spending. The internet was the gasoline on the fire.

Bubbles are critical for growth.

Real growth involves the generation of new material. This is true for our brains and the innovation-driven part of the economy. The bubble stage drives a massive wave of new investment that doesn’t demand a traditional “ROI” analysis. It’s like a gold rush where participants go “hell-bent for leather.”

Innovation and new product introductions go vertical. Some are winners, others are duds. Some “winners” lose money, but a few have economics that drive profits and an opportunity to invest more to consolidate a long-term winning position in the new market.

These are the ones that become Microsoft MSFT 0.00%↑, Amazon.com AMZN 0.00%↑, Uber UBER 0.00%↑ or AirBnB ABNB 0.00%↑ of the next wave of growth. We have Nvidia NVDA 0.00%↑ right now, and many other firms are vying for a piece of the pie.

Don’t sell a bubble short.

During a bubble, people constantly wonder, “When will it end?” The 1995-2000 bubble persisted for years. It’s impossible to know how long it will last; all one can do is follow it and be prepared.

Magnitude is an open question. Is “AI” more or less impactful than “The Internet?” I would say that AI is more significant and moving faster than internet technologies in the late 1990s.

The pace of AI advancement has been staggering in the past 12 months. The next release from OpenAI will continue this trend. It’s worth the 22 minutes it takes to absorb OpenAI o3 and o3-mini.

Investors are beginning to inquire about business models and returns, and efficiency is emerging as a primary consideration. These questions are unlikely to slow down investment in the near term.

Our data shows that enterprises have yet to invest seriously in AI technology. Most of what we’ve seen has been from hyperscalers. That means that despite the massive growth in 2024, most of the business world has yet to contribute.

The People Question

Once o3 and other advanced AI models are released, it begs the question of what people will do, particularly those you think of as “solid but uninspiring” professionals.

About 50 million jobs in the US will likely be eliminated by soon-to-be-current versions of AI. Investors may cheer the potential to increase operating margins (from historically high numbers). Still, they will also wonder about the impact of that level of job losses on consumer demand and the overall economy.

This brings up discussions of " universal basic Income,” which most people think will not solve real problems—people need something tangible to do.

My base case is that we’ve got years to play this out. But there will be plenty of uncertainty, which will breed volatility. That’s what I’m getting ready for.

As a reminder, this isn’t any flavor of investment advice but another stick on the campfire we’re all sitting around.