In November 2022, with FTX still smoking, I wrote a post called Exchanges Don’t Matter. Some people thought it was a hot take. Four years later, after the biggest theft in the history of money on the internet, I’m adding one word: they really don’t matter.
On the evening of October 29, 2026, roughly $6.36 billion left Binance hot wallets across ten blockchains, most of it in about 20 minutes (TRM Labs). That is about four times the Bybit hack of 2025. It happened to the largest, richest, most “secure” exchange on the planet.
Let's be clear about what happened: this wasn't a crypto failure. It was an exchange failure. And it may mark the moment the exchange era started to end.
After FTX, my argument was simple. The real innovation in crypto lives far beyond exchanges. Exchanges brought little financial or technological novelty. They mimicked a part-bank, part-brokerage, and did it badly on risk management and customer service.
They optimized for one thing, fast settlement of money transfers, without giving users direct ownership of their accounts. They were a low-level proxy to the blockchain, a temporary bridge for people who didn’t want the hassle of self-custody.
I also argued that a self-custodial wallet on a DEX can do everything a CEX does, and more. DEX failures are micro-managed through automatic liquidations, never systemic. Every “extra” exchanges added, from staking to loans to NFTs, followed what DeFi had already built. My conclusion: one day we’d realize exchanges were just a passage toward the decentralization of finance.
The Binance hack doesn’t revise that thesis. It confirms it, at 4x the scale.
The facts so far: blockchain analysts at TRM Labs and Chainalysis attribute the attack to TraderTraitor, a North Korean group. Binance first pointed to a compromised engineer’s laptop that let the attacker submit unauthorized withdrawals; later updates mention additional points of breach. Binance suspended deposits and withdrawals, halted trading, and pledged to cover all losses (Axios). About $3 billion has been frozen on-chain, much of it by stablecoin issuers like Tether and Circle (CNBC).
Notice what failed. Not Ethereum. Not Bitcoin. Not a single blockchain was broken. What failed was a company’s internal signing infrastructure, the one place where billions of other people’s coins sat behind one corporate perimeter.
Binance was too big, and that is exactly why it was targeted. A state-sponsored hacking team doesn’t spend months planning to rob 12,000 wallets of $600,000 each. It robs the one honeypot holding all of it. Concentration isn’t a side effect of the exchange model. It is the model. And concentration is what attracts attackers.
Central exchanges were created as a convenience. They were not part of the original vision. Read the Bitcoin white paper: its whole point was peer-to-peer value transfer without a trusted intermediary. A custodial exchange is another "trusted" intermediary with a ticker tape.
Anyone who was in crypto early enough will remember that exchanges were despised. We saw them for what they were: concentrated risk dressed up as access. Mt. Gox taught us that lesson in 2014. Quadriga's fiasco repeated in 2017. FTX taught it again in 2022. Bybit's hack happened in 2025. Binance just taught us the same lesson a fifth time, at the largest scale yet.
How many times does a lesson need to be repeated before we accept it?
There is nothing an exchange provides that you can’t do yourself with a bit of self-study. Buying, swapping, lending, borrowing, earning yield, bridging across chains: all of it is available through DEXes, DeFi and other decentralized services, with your keys in your hands.
In fact, you can do more outside exchanges. You can access assets an exchange hasn’t listed. You can compose services an exchange would never bundle. You can verify every rule onchain instead of trusting a terms-of-service page.
Yes, self-custody takes learning. So did online banking. So did the web. The learning curve is not an argument for handing your assets to a honeypot; it’s an argument for better wallets and better education. The industry should invest there, not in more glass towers. User experiences and security soundness must be improved in the self-custody side of the industry, because they still suck for the average consumer who are too lazy to jump the usage barrier, and tend to gravitate toward central exchanges for convenience, with a hail Mary on future safety.
The real tension in crypto has always been between two goals: (a) making crypto mainstream, and (b) staying true to the original vision of self-sovereign, disintermediated value.
Exchanges delivered (a). They onboarded hundreds of millions of people. Thank you very much for helping. But they did it while violating (b), re-creating the very custodial middlemen crypto was invented to remove. That trade-off was always going to come due. On 10/29, it did.
This is not unique to crypto. Every revolution gets hijacked somewhere along its history. The internet’s open protocols were captured by a handful of platforms. Then the pendulum swings, and the revolution gets back on track. Crypto’s hijacking had a name, and it was centralized exchanges.
If you’re watching from the outside, you have every right to be skeptical. Six billion dollars vanishing in an evening should make anyone pause.
But point your skepticism at the right target. The knock is not against the industry. It’s against the exchanges and their model. The blockchains did exactly what they were designed to do: they executed validly signed transactions, transparently, in public, where every analyst on earth could trace them in real time. The weak link was a corporate laptop and a centralized vault.
Ironically, the stablecoin freezes that clawed back about half the loot are themselves a reminder of how much centralization still sits in the stack. That’s a debate for another day. The point stands: blaming crypto for the Binance hack is like blaming the internet for a bank’s data breach.
I’ll say it plainly: Binance’s best days are behind it. It will cover the losses, it will survive, and it will spend years rebuilding trust it may never fully regain. More importantly, we may have just reached peak-exchange mode for the whole industry.
This is crypto’s 9/11 moment. Not in human cost, obviously, but in how it rewires behavior and assumptions. After 9/11, the way we traveled, secured buildings, and thought about risk changed permanently. After 10/29, the way people hold crypto will change too. I don’t think we will ever see a theft at this scale again, because the honeypots will shrink.
People will realize that self-custody and smaller concentrations of crypto are the way to go. Spread risk. Hold your keys. Use exchanges, if at all, as an on-ramp and off-ramp, never as a vault.
We will remember this as Before 10/29 and After 10/29. And years from now, we may look back and conclude what I suggested in 2022: exchanges really didn’t matter. They were a detour. The road head was always decentralization.
References: