USPTO Patent Grant: 3D Hybrid Bonding Memory Devices for AI Inference

ChangeBridge: Patent Grants - AI & Computing (G06N)

Published March 24th, 2026

Detected March 24th, 2026

Summary

The USPTO has granted a patent (US12585931B2) for 3D hybrid bonding memory devices designed for AI inference applications. The patent, assigned to Macronix International Co., Ltd., details a platform integrating logic and memory dies with specialized AI processing elements.

View original document View source feed page

What changed

The United States Patent and Trademark Office (USPTO) has issued patent US12585931B2, titled '3D hybrid bonding memory devices with NPU/CPU for AI inference application'. This patent, granted to Macronix International Co., Ltd., describes an AI inference platform that features a logic die with AI processing elements and a memory die with 3D memory cells. The innovation lies in the direct vertical connections between AI processing elements and memory buffers, enabling efficient storage and retrieval of neural network weights for AI computations.

This patent grant is primarily an intellectual property development and does not impose new regulatory obligations on businesses. Companies involved in AI hardware development, semiconductor manufacturing, or AI inference platforms may find the technical details of this patent relevant for their own research and development strategies. No compliance actions or deadlines are associated with this patent grant.

Source document (simplified)

← USPTO Patent Grants

3D hybrid bonding 3D memory devices with NPU/CPU for AI inference application

Grant US12585931B2 Kind: B2 Mar 24, 2026

Assignee

MACRONIX INTERNATIONAL CO., LTD.

Inventors

Hsiang-Lan Lung

Abstract

An AI inference platform comprises a logic die including an array of AI processing elements. Each AI processing element including an activation memory storing activation data for use in neural network computations. The platform includes a memory die that includes an array of 3D memory cells and a page buffer that facilitates storage and retrieval of neural network weights for use in neural network computations. A plurality of vertical connections can directly connect AI processing elements in the logic die and page buffers of corresponding ones of the memory cells in the memory die, enabling storage or retrieval of a neural network weight to and from a particular page buffer of a corresponding 3D memory cell for use in neural network computations conducted by a corresponding AI processing element in the logic die.