Brave New World

In the previous post, we looked behind the AI frenzy at the mounting debt towers and the zero-sum game of chicken playing out across the market.

Facing that diagnosis, a familiar dread creeps in:

"Is it all over, then?
When the bubble pops, don't AI and startups all go down together?"

People said the exact same thing when the dot-com bubble burst in 2001. Driven by the conviction that the internet would change the world, thousands of dot-coms flooded the market, only to shutter overnight once the mania cooled. Silicon Valley became a wasteland, and skeptics wrote off the entire web as the greatest scam in history.

Yet only a few years later, something strange happened.

The Paradox of the Boom

Why did the true golden age of the internet emerge from the rubble of a crash, rather than the peak of the cash-flush bubble?

Founders often assume boom times, overflowing with capital, make the best season to start a company. That is only half right. Raising venture money might be easy; building a durable business is anything but.

When money is cheap, hollow companies crowd the field. Lacking clear business models or real tech, they raise checks by hanging trendy buzzwords on their shingles. Flush with investor cash, they subsidize freebies to inflate headcount and user vanity metrics, masquerading as innovators.

When zombie companies* run rampant, teams trying to build on real fundamentals suffocate. Cheap capital drives engineering salaries and customer acquisition costs through the roof. Subsidized freebies blind customers, making it impossible to separate real value from venture-funded handouts.

Every market needs a cold shower once in a while.[1]

When the bubble deflates and liquidity dries up, the pretenders living on venture life support collapse. Inflated payrolls and ad rates reset to sanity, clearing out the noise polluting the market.

A downturn is not a catastrophe that ruins everything. It is capitalism's winnowing process—leveling an overheated playing field to separate the chaff from the operators producing tangible value.

The Rails Left Behind

Once the dust settles and the pretenders clear out, the real story begins.

This is where capitalism pulls off its best plot twist: speculators and shell companies go bankrupt, but the infrastructure they built stays behind.[2]

During Britain's Railway Mania in the 1840s, speculators tore across the country laying tracks in a fever to strike it rich. When the bubble burst, railway operators went under en masse. But the tracks stayed in the ground. That overbuilt network became the circulatory system of the British Industrial Revolution.

Today's AI arms race among Big Tech players follows the exact same path.

Desperate to protect their moats, tech giants bleed cash to distribute models. Even OpenAI admits that heavy users on its $200-a-month tier run at a net loss[4]. Yet to outmuscle rivals, they have slashed token pricing by more than 90% and flooded the web with subsidized access.

When this bubble pops, venture portfolios will take a beating, but the assets funded by Big Tech's tuition will not vanish: open-source weights anyone can run, dirt-cheap compute, and hundreds of millions of users who now treat AI as a baseline habit.

Big Tech burned their own balance sheets to pave a high-speed highway right in front of us.

Brave New World

Once the highway opens and tolls drop to near zero, the real protagonists step onto the stage*.

Take the 2001 dot-com crash. Telecom giants borrowed hundreds of billions to bury fiber-optic cables under oceans, banking on an internet traffic boom, and went belly-up. But the dark fiber they left behind for pennies on the dollar made bandwidth essentially free—setting the stage for Amazon, Google, Netflix, and the modern cloud.[3]

"So, will the AI bubble burst?"

Draw your own conclusions.

What is certain: on the other side of the pop lies a mountain of discounted tech. Founders will pick up those cheap modular parts to tackle urgent, real-world problems—problems painful enough that customers gladly pay full price, without discounts or subsidies.

Investors dread the coming winter.

For builders, a change of season is nothing to fear. Hype clears, tech turns into a cheap commodity, and the playing field levels out.

The brave new world for real builders starts in the rubble.

(End.)

References

1. Cleansing vs. Scarring Effects of Recessions:
   Economist Joseph Schumpeter argued that downturns serve a cleansing function, liquidating inefficient, marginal firms and reallocating resources to true innovators. Modern macroeconomic studies note that severe credit crunches can also inflict scarring effects, taking down promising early-stage companies starved of bridge capital. In a downturn, internal cash generation without external funding becomes the ultimate litmus test for survival.
2. Technological Revolutions and Financial Capital:
   Formulated by Venezuelan scholar Carlota Perez. Analyzing five major technological revolutions over the past 250 years (the Industrial Revolution, Steam and Railways, Steel and Heavy Engineering, Oil and the Automobile, Information and Telecommunications), Perez showed that every technological shift runs on two distinct cycles: 1) the Installation Period, where financial capital inflates a speculative bubble to overbuild infrastructure, and 2) the Deployment Period, where after the crash, cheap infrastructure diffuses across the real economy to generate productive growth.
3. The Legacy of the Dot-Com Crash:
   In the late 1990s, telecom companies spent over $500 billion laying global fiber networks. When the bubble burst in 2001, giants like WorldCom and Global Crossing went bankrupt, leaving over 90% of global dark fiber unused. That glut crashed bandwidth costs, paving the foundation for Amazon, Google, YouTube, Netflix, and the hyperscale cloud.
4. Subsidized Burn in AI Labs:
   Despite billions in top-line run rates, labs like OpenAI and Anthropic burn billions in investor capital on training runs and subsidized inference. OpenAI CEO Sam Altman noted in early 2025 that power users on the $200-a-month plan consume so much compute that the company loses money with every heavy query.

Key Terms

• Zombie Company:
  A marginal firm that cannot cover its debt servicing costs from operations, surviving only on low-interest debt, government handouts, or constant venture capital infusions.
• Installation vs. Deployment Periods:
  1) Installation Period: Financial capital floods into new technology, driving reckless infrastructure buildouts, asset bubbles, and wealth polarization.
  2) Deployment Period: After the bubble bursts, the cheap, ubiquitous infrastructure integrates across traditional industries, generating tangible productivity and economic value.