My garbage can got repossessed in 2001.
I know that sounds absurd, but it perfectly captures the surreal devastation of the dot-com crash. One day I was a young engineer at a hot startup, stock options growing more valuable by the week, new house, first baby on the way. The next day the company vanished, my options were worthless, and we were losing everything—including our municipal garbage service.
For months afterward, I drove hours from home for consulting gigs, watching the wreckage of Silicon Valley’s first great bubble. Half the internet companies disappeared within two years. 48% of them, gone. Along with hundreds of thousands of jobs and families like mine.
I thought we’d learned something from that carnage. I was wrong.
Last month, Builder.ai declared bankruptcy, ending the story of a company that raised $450 million and reached a $1.3 billion valuation by promising AI-powered app development. Turns out the AI was mostly just engineers in India doing manual coding. Bloomberg later revealed the company had inflated revenue figures by up to 300% through fake transactions with partner companies.
As I watched the news break, that familiar feeling washed over me. Not surprise—recognition. I’ve seen this movie before. Multiple times. The script never changes, just the technology and the names.
Act I: The Internet Gold Rush (1995-2001)
The template for all future bubbles was written during the internet boom. Between 1995 and 2000, the NASDAQ surged 582% as venture capitalists poured $120 billion into internet startups. Companies like Pets.com achieved $300 million valuations despite having no path to profitability.
The mantra was intoxicating: get big fast, figure out profits later. 72% of 1999 IPOs had negative earnings, yet they traded at extreme valuations because “eyeballs” and “click-through rates” mattered more than revenue.
I was part of that fever dream, believing we were building the future while $1 trillion flowed into fiber-optic networks and data centers that nobody actually needed yet. Enough dark fiber was laid to circle the globe 1,000 times. The infrastructure build-out was so extreme it took a decade to absorb the excess capacity.
The crash was swift and merciless. $5 trillion in market value evaporated by 2002. Survivors like Amazon required 15 years to regain their peak valuations. The human cost was staggering—not just paper losses, but real families losing real homes.
But here’s the crucial detail: the crash didn’t kill the internet. It just reset expectations. The overbuild infrastructure became the foundation for Google, Facebook, and Web 2.0. The boom was wrong about timing and business models, but right about the technology’s transformative potential.
That paradox—waste and innovation intertwined—would define every bubble that followed.
Act II: Clean Energy Dreams (2006-2013)
By 2006, we were ready for the next revolutionary narrative. Climate change was becoming undeniable, oil prices were spiking, and venture capitalists found their new obsession: clean technology.
VCs poured $25 billion into clean energy startups, convinced they were funding the next industrial revolution. Solar panel manufacturers secured billions despite Chinese competition that was already driving down prices. Biofuel companies like KiOR raised enormous sums for technology that simply didn’t work at scale.
This bubble had a different character than the internet boom. These were capital-intensive hardware plays with 7-10 year development cycles that clashed with 3-5 year VC fund horizons. 68% of business models relied on government legislation that never materialized, creating regulatory risk that investors consistently underestimated.
The crash was brutal but predictable. Clean tech’s share of VC funding plummeted from 15% to 3% by 2013, with 90% of biofuel startups filing bankruptcy. Chinese manufacturing erased tens of billions in value through commoditization that VCs never saw coming.
Government interventions made everything worse. Stimulus funds created artificial demand that disappeared when subsidies expired. Companies that couldn’t survive without regulatory support were zombie firms from day one.
Yet again, the underlying technology survived and eventually thrived. Solar costs dropped 90% through manufacturing scale achieved during the bubble. The clean tech crash drove innovation toward software-enabled solutions and more sustainable business models.
The pattern was becoming clear: revolutionary technology plus unrealistic expectations equals spectacular waste—followed by genuine progress built on the wreckage.
Act III: The Theranos Playbook (2003-2018)
While clean tech was imploding, a different kind of company was perfecting a more sophisticated form of deception. Theranos didn’t just ride a bubble—it created a masterclass in how to sustain fiction through charisma, partnerships, and carefully managed opacity.
The company raised over $700 million claiming its proprietary Edison devices could run hundreds of blood tests from a single drop of blood. The reality? Most tests were run on traditional third-party machines, with Edison devices that barely worked.
But Theranos mastered something previous bubbles had stumbled upon accidentally: credibility by association. Henry Kissinger on the board. Partnerships with Walgreens and Safeway. Media coverage that asked few technical questions but celebrated the vision endlessly.
The Theranos playbook became a template:
- Make revolutionary technology claims
- Secure impressive partnerships and board members
- Maintain operational opacity while highlighting credentials
- Deflect technical scrutiny through narrative control
- Use each funding round as validation for the next
When the fraud was finally exposed, it revealed systematic deception that went far beyond normal startup optimism. But by then, the playbook was already being copied across Silicon Valley.
Act IV: Crypto Casino (2017-2023)
The cryptocurrency bubble was different—bigger, wilder, and more democratized than anything before it. When retail investors could participate directly through Coinbase and Robinhood, speculation became a global phenomenon.
In 2017, companies began raising billions through Initial Coin Offerings with whitepapers that read like science fiction. Most projects had no working product, no revenue, no clear path to either. But they had compelling narratives about the future of money and 58 million global participants willing to bet on those stories.
The 2021 NFT mania pushed speculation to absurd extremes. Profile picture collections generated $33 billion in trading volume, with digital images selling for hundreds of thousands of dollars. The underlying blockchain technology was interesting, but the valuations were completely detached from utility.
Even as the market was collapsing, VCs invested $52 billion in crypto projects between 2021 and 2022. They weren’t stupid—they were following the playbook, betting on momentum rather than fundamentals.
The unraveling was spectacular. Terra/Luna’s $40 billion collapse from flawed algorithmic design. FTX’s $8 billion in missing customer funds. By 2023, 94% of NFT collections had become illiquid, with median prices down 98%.
Unlike previous crashes, crypto left behind a more ambiguous legacy. Some infrastructure innovations were genuine—programmable money, decentralized finance, digital ownership. But much of the “innovation” was financial engineering designed to extract value rather than create it.
Act V: The AI Reckoning (2020-Present)
Now we’re watching the same script play out with artificial intelligence, and it’s the most frustrating one yet because the underlying technology is genuinely revolutionary.
Generative AI funding reached $56 billion in 2024, representing a 92% increase year-over-year. Tech giants have committed hundreds of billions for AI infrastructure. It’s the biggest gold rush in Silicon Valley history.
But here’s what drives me crazy: 78% of AI startups lack clear monetization paths beyond API access. They’re bullshit companies with bullshit business plans, founded by people who have no intention of building sustainable businesses. They’re designed for fast exits—IPOs or acquisitions—before anyone notices the emperor has no clothes.
The evidence is mounting everywhere you look. Research from Orgvue shows that 55% of companies that replaced humans with AI now regret those decisions. IBM spent over $200 million automating HR functions, only to hire back “even more” people than they initially laid off when the AI couldn’t handle complex human interactions. McDonald’s AI drive-thru became a viral embarrassment, confidently taking orders for “260 McNuggets with ketchup” before the company quietly abandoned the technology.
Duolingo’s “AI-first” strategy backfired so spectacularly that they had to delete all content from 6.7 million TikTok and 4.1 million Instagram followers when users revolted against the CEO’s tone-deaf proclamations that AI would soon teach better than humans. These aren’t isolated incidents—they’re symptoms of an industry that’s confused hype with strategy.
Builder.ai was the perfect example. For years after being exposed in 2019 for using human engineers instead of AI automation, the company kept raising money. Microsoft invested. Qatar Investment Authority invested. The founder won Ernst & Young’s UK Entrepreneur of the Year award.
The warning signs were everywhere, just like they always are. Revenue inflated by up to 300% through fake transactions. Operational claims that didn’t match reality. A business model that required continuous funding to survive.
When Builder.ai collapsed last month, taking 270 jobs with it, it felt like watching my dot-com startup disappear all over again. Same script, different decade.
Why the Script Never Changes
After living through four major bubbles, I’ve realized something uncomfortable: these cycles aren’t accidents. They’re features of how venture capital works.
The math drives everything. VC funds need to deploy capital quickly to meet pacing requirements. When hot sectors emerge, funds can’t afford to sit on the sidelines while competitors write checks. The result is systematic momentum chasing rather than careful evaluation.
Fund structures make it worse. VCs get evaluated on paper valuations in the short term, not actual cash returns. A partner who leads a Series A that doubles on paper gets credit for success, even if the company eventually goes bankrupt. Early paper returns are prioritized over distributions, enabling valuation inflation across entire sectors.
Information asymmetry compounds the problem. Most VCs don’t have deep technical expertise in the sectors they fund. They rely on pattern matching and social proof. When Andreessen Horowitz leads a round, suddenly every other firm wants in, regardless of fundamentals.
The result is predictable: 78% of 2021 crypto investments came from generalist funds rather than sector specialists. They were betting on hype, not knowledge.
The Coming AI Correction
Unlike crypto or clean tech, I’m genuinely bullish on AI’s long-term prospects. The technology is transformative. The applications are real. The potential is staggering.
But that doesn’t mean we won’t see a correction that separates legitimate companies from exit-focused hype merchants. As I’ve written before, the choices leaders make in the next 18 months will determine whether AI creates the greatest economic expansion in human history or triggers a cascade of market failures that makes previous crashes look manageable.
The correction is coming not because AI is overhyped, but because most AI companies are. We’re watching what economists call a Nash Equilibrium trap play out in real time—individually rational decisions (replace expensive humans with efficient AI) that become collectively catastrophic when everyone makes the same choice.
When it arrives, it won’t destroy the sector the way crypto winters devastated blockchain projects. Instead, it will perform a necessary function: eliminating companies that exist solely to ride the hype wave to acquisition.
The real tragedy is how these bullshit AI companies are poisoning the well for everyone else. They’re creating anti-AI sentiment in society and skepticism in the tech community. When people see AI companies making grand claims without substance—or worse, eliminating jobs for quarterly gains while delivering inferior results—it breeds cynicism about the entire field.
After the correction, we’ll finally separate signal from noise. The infrastructure investments from the bubble will become cheaper and more accessible. The conversation will shift from “AI will replace everything” to “AI will enhance specific workflows in measurable ways.”
That’s when real innovation happens—when technology matures beyond hype and starts solving actual problems profitably. The companies that survive will be those that figured out how to amplify human potential rather than replace it, building sustainable competitive advantages through human-AI collaboration instead of chasing quick exits through automation theater.
Breaking the Pattern
The question isn’t whether these cycles will continue—they will. The incentive structures are too entrenched and the psychology too hardwired. The question is whether we can reduce their destructive impact.
Some changes are already emerging. 32% of new VC funds now tie compensation to actual cash distributions rather than paper valuations. Others require specialized investment theses instead of generalist momentum chasing.
Technology is also democratizing innovation. Open source AI frameworks are reducing development costs dramatically, enabling smaller bets on more diverse approaches. Maybe we don’t need unicorns at all—maybe we need hundreds of specialized companies carving out profitable niches.
But the most important change needs to happen in how we think about success. Instead of swinging for trillion-dollar companies that justify 90% failure rates, what if we supported sustainable businesses that solve real problems for real customers?
The obsession with finding “the next Google” creates these destructive cycles in the first place. Tech giants have grown so large they cast shadows that prevent anything else from growing. A more distributed model would mean lower failure costs, more innovation diversity, and reduced systemic risk.
The View from the Wreckage
I’ve now watched the same movie play out four times: dot-com, clean tech, crypto, and the beginning of the AI crash. Each time, I hoped we’d learned something from the previous carnage. Each time, I was disappointed.
But maybe that’s the wrong way to think about it. Maybe these cycles aren’t bugs in the system—they’re how the system works. Venture capital funds genuine innovation by also funding spectacular waste. The infrastructure investments during bubbles become the foundation for the next wave of progress.
The dot-com overbuild enabled Web 2.0. Clean tech manufacturing scale drove down solar costs. Crypto speculation advanced digital payments. AI infrastructure will enable applications we can’t yet imagine.
The human cost remains real and devastating. Every bubble collapse means families losing homes, careers getting derailed, communities losing economic anchors. That garbage can repossession wasn’t just absurd—it was a symbol of how financial abstraction translates into personal catastrophe.
But the alternative—slow, careful, conservative investment—might mean missing transformative technologies entirely. The venture capital model, for all its flaws, has funded most of the innovation that defines modern life.
As I watch the AI bubble inflate, I feel that familiar mix of excitement and dread. The technology will change everything. Most of the companies will disappear. Families will get disrupted. Infrastructure will get built. Progress will emerge from the wreckage.
The script never changes, but maybe that’s because the script works—just not for everyone, and not without tremendous cost.
The next time someone tells you they’ve found a revolutionary technology that will change everything, ask yourself: have I seen this movie before? Because if you’ve been paying attention, you probably have.
The ending is always the same. Only the popcorn flavors change.