Future of Computation Is Encrypted

Your data is encrypted when it’s stored. It’s encrypted when it’s sent. But the moment it’s used? It’s exposed.

That’s the flaw behind most data breaches you’ve heard of.

Fully Homomorphic Encryption (FHE) fixes that. It lets third-parties perform computations on encrypted data — without ever decrypting it. The server never see your input nor the output. Never touches plaintext. And yet, you get the correct result.

Imagine asking ChatGPT a sensitive question. It gives you the right answer — but never saw your question. Sounds impossible. But it’s possible. Though Its high computational overhead is preventing it to be practical.

That being said, there are already practical applications of it. Apple uses FHE for private ML. Microsoft uses it for password safety. Researchers use it for encrypted genome analysis.

The first FHE systems were painfully slow. But things are changing fast. Core algorithms are improving ~8× each year. FHE-specific hardware is getting better.

With this pace of progress, FHE could become feasible in many more applications soon.

And when that happens, it will change everything. A new internet will be built — one that doesn’t spy by default, but private by default.

Private LLMs, confidential blockchains, and searchable encrypted databases might become the norm.

Startups are building confidential blockchains, private LLMs, and searchable encrypted databases.

The future computation is encrypted.