I wrote Blockchain For Dummies in 2017. I've watched every wave of blockchain hype from close enough to see what was real.
Here's what I've concluded after nearly a decade: the technology was never the problem. The business model usually was.
What Didn't Work (and Why)
ICOs — Initial Coin Offerings were a fundraising mechanism masquerading as a product launch. Most projects raised capital on a whitepaper, spent it building something nobody used, and disappeared. The ones that survived either found genuine utility or pivoted to infrastructure.
Consumer NFT speculation — Buying a JPEG and hoping it appreciates isn't a business model. The projects that framed NFTs as speculative assets failed. The ones that used them as access tokens, membership mechanisms, or provenance systems had more staying power—but even most of those are gone now.
DeFi yield farming — Yielding 400% APY on a token that depreciates 99% isn't yield. It's redistribution. Real DeFi—automated market makers, overcollateralized lending, decentralized stablecoins—has real utility. The farms built to extract from that utility didn't.
Crypto payments at retail — Volatility and UX killed this. Consumers don't want to think about gas fees at checkout. This wasn't a technology problem; it was a product problem from the start.
What Actually Worked
Stablecoins — USDC, USDT, and the handful of overcollateralized alternatives have genuine product-market fit. Cross-border payments, DeFi rails, emerging market savings. The market tells you what's real: stablecoin transaction volume consistently eclipses speculative token volume.
Tokenization of real assets — Private credit, real estate, treasuries, carbon credits. Bringing off-chain assets on-chain with legal enforceability solves a real problem: illiquidity in private markets. BlackRock isn't tokenizing treasuries because blockchain is interesting. They're doing it because settlement is faster and the audit trail is cleaner.
Blockchain for supply chain provenance — Not as a consumer feature, but as a B2B compliance mechanism. Food safety, pharmaceutical track-and-trace, luxury goods authentication. These are boring use cases. They work exactly because they're boring—they solve a real audit and liability problem.
Layer 2 infrastructure — The rollup ecosystem is genuine engineering progress. Lower fees, faster finality, Ethereum security. This is what enables the other categories above to actually function at scale.
Crypto-native financial services — Wallets, custody, on/off ramps, exchanges. Not glamorous. Essential infrastructure for anyone building in the space.
The Pattern
The blockchain business models that survived share one thing: they replaced something that was worse before, and the blockchain component was the minimum viable change to make the replacement work.
Stablecoins replaced wire transfers for cross-border settlement. Tokenization replaced paper processes in private markets. Supply chain provenance replaced spreadsheets and phone calls.
The projects that failed usually added blockchain to something that didn't need it, then tried to justify the complexity with a token.
Where This Goes
The next wave is AI + blockchain—and I'm watching it carefully. There are real use cases: verifiable computation, provenance for AI-generated content, decentralized training markets. There's also a lot of AI-washed blockchain theater that looks exactly like ICO theater did in 2017.
The filter is the same: is the blockchain component doing real work that couldn't be done better another way? If the answer is yes, it's probably interesting. If the answer requires a whitepaper to explain, be skeptical.
I've spent a decade being early to this technology and watching it grow into genuine infrastructure. The hype cycles are frustrating. The underlying innovation is real.