Claude isn’t just helping people write code, but it’s also leading to some windfall gains.
A user on X going by the handle @cprkrn posted what may be the most dramatic AI success story of the year: Anthropic’s Claude helped him recover 5 Bitcoin — worth roughly $395,000 at the time of recovery — that had been locked away in an old wallet since 2013.
The backstory is equal parts relatable and painful. Back in college around 2013, the user bought Bitcoin at around $200 a coin at a Starbucks, paying cash through LocalBitcoins. He then made a password for his blockchain.info wallet that he later couldn’t fully recall. He had two of the three components of the password but couldn’t crack the third. “Kinda like a forced hold,” he said in an interview with MTS. “I was like, ‘You know what? Hey, I would’ve sold this thing anyway. Let’s just like wait till it’s a hundred K or a little more.'”
Back in 2023, he’d posted how he had been locked out of his account. “My wallet with my locked btc from 9 years ago lol,” he’d posted, with a link to a wallet showing 5 bitcoin.
Over the years, he tried everything — “trillions of passwords,” as he put it, through conventional tools — with no luck. Then came what he called a last-ditch effort. He went to his parents’ house, dug through old college notebooks in the shed, and started feeding anything that looked like a password or seed phrase into Claude. That session cracked one wallet. But there was still another.
The real breakthrough came the morning after. He had one remaining laptop from his college days he hadn’t gone through. He opened Claude, dumped the files from that laptop into it, and let the model work. Claude found an old wallet file, connected it to the seed phrase from the previous session, and decrypted it. The 5 BTC was recovered.
“Something I just never would’ve tried on my own,” he said.
His post on X went viral, tagging both @AnthropicAI and CEO @DarioAmodei, with the user adding, only half-jokingly, that he’d be naming his kid after Amodei.
The story is a vivid illustration of how AI’s practical utility extends well beyond its headline use cases. While coding and developer workflows have grabbed the most attention — Claude Code crossed $2.5 billion in annualized run-rate revenue earlier this year — the model’s ability to reason through messy, unstructured real-world problems is proving just as valuable in unexpected contexts.
What makes this case technically interesting is the approach. Rather than brute-forcing through password combinations, Claude apparently reasoned through fragmented, context-rich inputs — notebook scribbles, old files, partial phrases — and connected the dots across sessions to reconstruct what the user’s college-era self would have chosen. It’s the kind of associative, pattern-matching work that suits a language model far better than conventional cryptographic tools.
Anthropic’s Claude has seen its traffic grow roughly 5x over the past year, driven in large part by an expanding base of users discovering new uses for the model that go beyond writing and code. From fraud detection systems built in 30 minutes to a guy recovering a forgotten crypto fortune from a college laptop, the pattern is the same: Claude functioning as a capable reasoning partner on tasks that have no obvious off-the-shelf solution.
For this particular user, the return on a $200 investment made at a Starbucks in 2013 came out to roughly $395,000 — and all it took was a decade of waiting, a trip to his parents’ shed, and an afternoon with an AI.