Intel · Filed Mar 27, 2026 · Published Aug 6, 2026 · verified — real USPTO data

Intel Patents Memory That Erases Itself After Being Read Once

Intel is patenting a memory chip that can be told to forget data the moment it's been read. It's a hardware-level take on the kind of 'read and destroy' logic that security engineers usually have to build in software.

Intel Patent: Read-Once Memory for Data Freshness — figure from US 2026/0228132 A1
Figure from the official USPTO publication.
See all 5 drawings from this filing ↓
Publication number US 2026/0228132 A1
Applicant Intel Corporation
Filing date Mar 27, 2026
Publication date Aug 6, 2026
Inventors Scott CONSTABLE, Simon P. JOHNSON, Fangfei LIU, Raghunandan MAKARAM, Thomas UNTERLUGGAUER
CPC classification 711/118
Grant likelihood Medium
Examiner CENTRAL, DOCKET (Art Unit OPAP)
Status Prosecution Suspended/Delayed (Jun 10, 2026)
Document 20 claims

What Intel's self-erasing memory chip actually does

Imagine a sticky note that vanishes the instant someone reads it. That's essentially what Intel is building into computer memory here. Certain pieces of data can be flagged so that once a program reads them, they're gone from memory automatically.

This matters because sensitive data, things like encryption keys or one-time tokens, often linger in memory long after they've served their purpose. Normally, a programmer has to write extra code to clean that up. Intel's design bakes the cleanup directly into the memory chip itself.

The chip uses a small marker called a read-once indicator attached to specific data. When anything tries to read that data, the chip hands it over and then immediately makes it unavailable for any future read. No extra software step required.

How the read-once indicator triggers data erasure

The patent describes a memory device with two main parts: a standard memory array that holds data, and a control circuit that watches over how that data is accessed.

The key feature is a per-data flag called a read-once indicator. When this flag is set on a piece of stored data, the control circuit enforces a strict rule: the data can be read exactly once. On that read, the control circuit does two things simultaneously: it outputs the data to whoever requested it, and it renders the data unavailable for any future read operation.

  • Memory array: stores data normally, with optional read-once flags attached
  • Control circuit: intercepts read requests and checks whether the read-once flag is set
  • Erasure action: immediately makes the data inaccessible after the first successful read

The patent frames this under "memory freshness," a concept in secure computing that ensures a processor can trust that the data it receives hasn't been replayed or reused from a stale copy. By making the hardware enforce single-use data, the design removes a whole class of software bugs where sensitive values stick around in memory longer than intended.

We find one patent like this every day. Get the best of each week in your inbox, free →

What this means for chip-level data security

For security-sensitive chips, things like secure enclaves, trusted execution environments, or cryptographic accelerators, having the memory hardware enforce single-use data is a meaningful upgrade. It removes the risk of a software bug or a timing attack allowing an attacker to read a sensitive value twice before the software gets around to clearing it.

This fits Intel's ongoing work on confidential computing, where the goal is to protect data even from a compromised operating system or hypervisor. If your encryption key literally cannot be read a second time at the hardware level, an attacker who gains software-level access after the fact finds nothing there.

Editorial take

This is a focused, practical security patent rather than a big-picture research idea. The concept of read-once memory exists in academic literature and niche secure hardware, but putting it into a standard memory device's control circuit is a real engineering step. It's worth watching because Intel has the manufacturing scale to ship this in production silicon, not just a research prototype.

There are more where this came from

We read every patent application Big Tech publishes and send you the ones worth knowing. Plain English, free, every week.

The drawings

5 drawing sheets from US 2026/0228132 A1 · click any drawing to enlarge

Patent filing page

Source. Full patent text and figures from the official USPTO publication PDF.

Editorial commentary on a publicly published patent application. Not legal advice.