Bubblicious

Build, Baby, Build

Over the last twelve months, the AI arms race has started to resemble something between the Space Race and the California Gold Rush.

The world’s largest technology companies are on pace to spend roughly $650 billion this year building AI infrastructure. Data centers are being constructed so quickly that power has become the industry’s primary bottleneck. Utilities are restarting nuclear power plants. Peter Thiel recently led a$140 millioninvestment into Panthalassa, a startup that wants to build floating AI data centers in the middle of the ocean, cooled by seawater and powered by wave energy. Elon thinks we’ve already outgrown Earth altogether. His long-term vision is to launch solar-powered AI data centers into orbit.

Investors seem perfectly happy to go along with it. The stock market just posted its strongest quarter in roughly six years. Every earnings call features another staggering capital expenditure budget. Every week brings another announcement that would have sounded completely absurd two years ago.

The reaction is almost universal.

“This is a bubble.”

Depending on who’s saying it, the word is delivered with concern, amusement, or a certain smug satisfaction. Calling something a bubble has become one of our favorite intellectual habits because it allows us to stand outside the crowd. Everyone else is caught up in the excitement. We alone are thinking clearly. Skepticism has become synonymous with intelligence while enthusiasm is treated as evidence of naivety.

The question isn’t if it’s a bubble. Of course it is. That’s a good thing.

Bubbles are good.

Exit Planet Dust

Venture capital does not generate returns by collecting dividends or patiently harvesting cash over decades. It generates returns when somebody is willing to buy what you have built. Historically, that has meant M&A, IPOs, and, recently, secondary transactions. Without those liquidity windows, venture capital may own extraordinary companies while returning very little actual cash to investors.

Perhaps, not surprisingly then, bubbles are the period when the majority of venture investments turn into that cash. From 1980 through 2025, VC-backed IPOs created roughly $3.47 trillion of market capitalization with more than half of that value originating from just five exuberant years: 1999, 2000, 2019, 2020, and 2021. Those years represented roughly 23% of VC-backed IPOs but about 56% of VC-backed IPO market cap. Add 2025 as the beginning of the AI liquidity window and six years account for nearly two-thirds of all VC-backed IPO market cap since 1980.

Cambridge Associates tells the same story: the two strongest calendar years in the history of its U.S. Venture Capital Index were 1999 and 2021, the two years most associated with speculative technology manias. Venture capital’s greatest returns are not evenly distributed across history. They concentrate during moments when markets are unusually willing to believe in the future.

Debbie Downers

History has a funny way of rewarding the people who build the future while celebrating the people who predicted its collapse.

Every generation tells the same story. Prices become detached from reality. Investors lose discipline. Companies are funded that never should have existed. Eventually the music stops, fortunes disappear, and everyone congratulates themselves for recognizing the excess. It’s a satisfying narrative because it reinforces the comforting belief that discipline always triumphs over exuberance and that skepticism is simply another word for intelligence.

Our culture compounds the problem by assigning status to pessimism. Calling something a bubble signals sophistication. It suggests you’ve maintained your objectivity while everyone else has become intoxicated by the moment. Optimism, by contrast, feels unsophisticated. It sounds promotional, self-interested, even gullible. We instinctively trust the person explaining why something won’t work more than the person imagining how it might.

The problem is that this is only half the story.

The other half is considerably more interesting. While investors don’t always survive the bubble, civilization has an uncanny habit of keeping everything the bubble built.

All Aboard

Britain experienced its own version in the 1840s during what became known as Railway Mania. Investors became convinced that railroads would fundamentally reshape commerce, transportation, and the economy. They were right. They also became convinced that nearly every proposed railway company would become wildly profitable. They were wrong.

Between 1844 and 1846, Parliament approved more than 8,000 miles of new railway, an extraordinary amount of infrastructure for a country roughly the size of Alabama. Thousands of companies were formed. Engineers, surveyors, lawyers, and bankers were pulled into what felt like an unstoppable national project.

The inevitable correction arrived with brutal force. Many railway companies failed before laying a single mile of track. Newspapers declared the episode proof that speculation had overwhelmed reason and that Britain had succumbed to collective madness.

And yet.

6,000 miles of railway were ultimately built. Britain inherited a national transportation network that dramatically lowered the cost of moving people, goods, and ideas. Economic historians now view the railway as one of the foundational technologies of the Industrial Revolution, permanently reshaping the British economy.

The Internet Was Built Twice

The dot-com era followed almost exactly the same script. We remember Pets.com, the sock puppet, the Super Bowl ads. What we forget is that the same speculative frenzy financed one of the largest communications infrastructure buildouts in history.

The numbers were astonishing. At the peak of the telecom boom, cumulative investment across the cycle ran over $500 billion. Companies like Global Crossing, WorldCom, Qwest, Level 3, XO Communications, and 360networks raced to lay fiber across continents and oceans believing internet traffic would grow forever. They were right about the direction and wrong about the timing.

The result: catastrophic overcapacity. By the early 2000s, it became obvious that the sector had massively overinvested, particularly in long-distance fiber optic cable. Stock prices collapsed, debt loads became unmanageable, and companies that had been treated as essential infrastructure ended up in bankruptcy.

But the wreckage left something behind. In 2004, the cost of bandwidth on long-distance routes had fallen by more than 90%. Access prices fell by half. The reason was simple: too much network, not enough traffic.

At the time, this looked like one of the great capital allocation mistakes of modern business history.

And yet.

The bubble effectively prepaid the next era of the internet. The companies that came after the crash inherited bandwidth that was dramatically cheaper. Google did not have to finance the entire global internet backbone before indexing the web. YouTube could decide streaming video was viable.

This is the thing about bubbles. The first generation get destroyed. The second generation get bargains. The third generation treats the infrastructure as ordinary. What looked like wasted capital in 2001 became cheap bandwidth in 2005, streaming video in 2007, cloud computing in the 2010s, and the operating surface of modern life shortly thereafter.

Fitter, Happier, More Productive

One of the strangest assumptions we make about bubbles is that the wealth they create disappears when the market turns. It does not. Much of it becomes real. And it finds its way back into new financing for new bets.

This process has a name: entrepreneurial recycling. Researchers use it to describe the way successful exits recycle wealth, talent, experience, and knowledge back into the startup ecosystem. A recent paper on unicorn exits found that an IPO led, on average, to two additional investments and another $13 million invested by each early investor.

And then there’s the talent. At least 150 companies have already been founded by SpaceX alumni, raising $12.6 billion and creating more than 8,000 jobs. As SpaceX continues to create thousands of new millionaires, the important story will be what those millionaires do next. Many will become founders, angels, LPs, advisors, and early employees at the next company that sounds impossible.

Successful bubbles create people with capital, confidence, networks, and scar tissue. Optimists with money who have already seen one impossible thing become real and are therefore willing to finance the next one.

Bubblicious

We have developed an odd relationship with optimism.

We instinctively admire the skeptic because skepticism sounds disciplined. We instinctively distrust enthusiasm because enthusiasm feels unserious. Calling something a bubble has become less of an observation than a way of signaling that we’re smarter than everyone else.

History suggests we should be a little more humble.

Most bubbles burst.

Many fortunes disappear.

Some companies deserve to fail.

But the infrastructure remains.

The talent remains.

The capital gets recycled.

The next generation gets built by people whose success was financed by the previous generation’s excess.

AI will be no different.

Twenty years from now, we’ll laugh at many of today’s valuations. We’ll wonder why anyone thought this company or that company was worth hundreds of billions of dollars.

We probably won’t be laughing at the intelligence infrastructure they built.

Bubbles are not evidence that capitalism has failed. They are evidence that optimism occasionally outruns reality. And they are necessary tools to move civilization forward.

So enjoy the ride.