WiMi Achieves Coexistence of Lightweight Design and High Performance by Efficiently Embedding Quantum Modules into U-Net

Rhea-AI Summary

WiMi (NASDAQ: WIMI) announced QB-Net, a hybrid quantum-classical deep learning approach that embeds a pluggable Quantum Bottleneck Module into the U-Net architecture. WiMi says QB-Net reduces the bottleneck layer parameter count by up to 30x while maintaining performance comparable to classical U-Net.

QB-Net encodes classical features into quantum states, applies parameterized quantum circuits with entanglement for feature transformation, then decodes measurements back into classical tensors. The module is designed for minimal parameters, trainability, and plug-and-play integration without changing U-Net structure or training paradigms.

Positive

• Parameter reduction of up to 30x at the U-Net bottleneck
• Claims of comparable performance to classical U-Net despite far fewer parameters
• Plug-and-play module: embeds without modifying U-Net architecture

Negative

• Quantum hardware limits: current devices cannot support large-scale quantum neural networks
• QB-Net is a module-only enhancement rather than a full quantum U-Net or Transformer

Key Figures

Parameter reduction up to 30 times fewer parameters QB-Net bottleneck layer versus classical U-Net bottleneck

Quantum circuit parameters tens to hundreds of adjustable parameters Quantum circuit versus hundreds of thousands in classical bottleneck

Classical parameters tens of thousands of parameters Classical network mapping task compared with a single quantum state

Qubits needed a few dozen qubits Single quantum state achieving comparable expressive power

Market Reality Check

$2.37 Last Close

Volume Volume 144,984 versus 20-day average 83,660 (relative volume 1.73x). high

Technical Price 2.37 trading below 200-day MA 4.01, reflecting a longer-term downtrend.

Peers on Argus 2 Up 2 Down

WIMI fell 3.27% while peers showed mixed moves: ACCS up 8.77%, MCTR up 0.65%, SWAG down 1.19%. Momentum scanner flags both up and down moves in related names, suggesting stock‑specific pressure rather than a clean sector‑wide trend.

Historical Context

Date	Event	Sentiment	Move	Catalyst
Dec 22	Quantum AI update	Positive	-1.4%	Announced next‑generation hybrid quantum neural network for image multi‑classification.
Dec 22	Quantum AI update	Positive	-1.4%	Repeat entry: hybrid quantum neural network structure announcement on same date.
Dec 04	Hybrid QCNN study	Positive	+1.9%	Unveiled hybrid quantum‑classical architecture to improve multi‑class image classification.
Dec 04	Hybrid QCNN study	Positive	+1.9%	Repeat entry: hybrid quantum‑classical learning architecture announcement.
Nov 20	QGAN architecture	Positive	-5.6%	Proposed dual‑discriminator hybrid quantum‑classical QGAN design for image generation.

Pattern Detected

Quantum/AI R&D announcements have produced mixed reactions, with several positive‑sounding updates followed by flat or negative next‑day moves.

Recent Company History

Over the past few months, WiMi has repeatedly highlighted advances in hybrid quantum‑classical AI. On Nov 20, 2025, it detailed a dual‑discriminator QGAN architecture, followed by hybrid quantum‑classical image classification work on Dec 4, 2025. Another hybrid quantum neural network structure was announced on Dec 22, 2025. Despite the technically ambitious nature of these updates, price reactions ranged from modest gains to notable declines, indicating that such R&D news alone has not consistently driven sustained upside.

Market Pulse Summary

This announcement showcases WiMi’s QB-Net, embedding quantum modules into U-Net to cut bottleneck parameters by up to 30x while preserving performance. It extends a series of hybrid quantum‑classical AI developments disclosed in late 2025. Investors may monitor whether these architectures progress from research towards commercial products, how they interact with existing business lines, and whether future filings or disclosures clarify revenue impact and deployment timelines.

Key Terms

quantum computing technical

"The core advantage of quantum computing lies in its ability to express..."

Quantum computing is a type of advanced technology that uses the principles of quantum physics to perform calculations much faster than traditional computers. It can process vast amounts of information simultaneously, potentially solving complex problems that are currently impossible or take too long with regular computers. For investors, this technology could lead to breakthroughs in areas like cryptography, data analysis, and optimization, impacting financial markets and security systems.

qubits technical

"...through the superposition states of qubits and perform linear operations..."

Qubits are the basic units of information in quantum computing, similar to how traditional computers use bits. Unlike regular bits that are either 0 or 1, qubits can represent both at the same time, allowing quantum computers to process complex problems much faster. This potential for unprecedented speed and power could transform industries, making qubits a key focus for investors interested in cutting-edge technology.

u-net technical

"...embedding lightweight quantum computing modules into the classical U-Net deep learning architecture..."

A U-Net is a type of computer program that learns to identify and outline structures inside images, most often used in medical scans to separate tissues, tumors, or other features from the background. Think of it as an automated “coloring inside the lines” tool that gets better with examples; for investors, U-Net-driven products can improve diagnostic accuracy, speed up workflows, reduce costs, and influence regulatory review and market adoption for imaging-dependent businesses.

quantum circuits technical

"...transformed through quantum circuits, it is possible to achieve equivalent capabilities..."

A quantum circuit is a planned sequence of operations that manipulates quantum bits (qubits), the basic units of quantum computing, to perform a calculation. Think of it as a recipe or wiring diagram that uses uniquely quantum behaviors—like being in multiple states at once—to solve problems that classical computers handle slowly or not at all. Investors care because advances in quantum circuits can unlock new commercial applications, disrupt encryption, and create opportunities for hardware, software, and service providers.

AI-generated analysis. Not financial advice.

BEIJING, Jan. 02, 2026 (GLOBE NEWSWIRE) -- WiMi Hologram Cloud Inc. (NASDAQ: WiMi) ("WiMi" or the "Company"), a leading global Hologram Augmented Reality ("AR") Technology provider, released a breakthrough achievement—a hybrid quantum-classical deep learning technology based on parameter-efficient quantum modules, QB-Net (Quantum Bottleneck Network). This technology achieves a major breakthrough by embedding lightweight quantum computing modules into the classical U-Net deep learning architecture, reducing the number of parameters in the bottleneck layer by up to 30 times while maintaining performance comparable to that of the classical U-Net. This research and development outcome not only demonstrates the cutting-edge potential of hybrid quantum-classical artificial intelligence but also provides a brand-new optimization paradigm for traditional deep learning architectures.
The core advantage of quantum computing lies in its ability to express high-dimensional information through the superposition states of qubits and perform linear operations in exponentially dimensional spaces, endowing it with expressive and transformative capabilities that surpass classical architectures. However, at the current stage, quantum hardware is still unable to support large-scale quantum neural networks or construct complete quantum U-Net or quantum Transformer.
Therefore, WiMi has taken a completely different path: instead of building fully quantized AI models, it constructs quantum enhancement modules.
This concept stems from a key observation: the bottleneck layer of deep networks is essentially a problem of high-density expression of high-dimensional features, while quantum states are naturally suited to express extremely high-dimensional vector spaces.
When a classical network requires tens of thousands of parameters to accomplish a mapping task, a single quantum state can theoretically achieve the same or even higher expressive power with only a few dozen qubits. This means that as long as classical features can be mapped into quantum states and transformed through quantum circuits, it is possible to achieve equivalent capabilities with extremely low parameter counts.
Based on this idea, WiMi designed a pluggable Quantum Bottleneck Module. This module takes minimal parameter count, structural stability, trainability, and the ability to be integrated into classical networks as its core objectives and has been embedded into the classical U-Net, forming QB-Net.
QB-Net retains the overall structure of U-Net, including the encoder, upsampling path, and skip connections. However, at the bottleneck layer position, the traditional multiple convolutional layers are replaced with a quantum feature compression-transformation-reconstruction module. This module consists of three key steps:
The first step is the encoding of classical features into quantum states. The encoding module uses techniques such as linear projection or amplitude encoding to map the classical feature tensor into a compact vector form suitable for entering quantum circuits. The design of the encoding strategy follows two major principles: minimizing the number of qubits as much as possible while preserving the key information of the features without loss.
The second step is feature transformation through quantum circuits, which is the core link of the entire system and the key to parameter efficiency. A traditional convolutional bottleneck layer may contain hundreds of thousands or even millions of parameters, whereas a quantum circuit requires only tens to hundreds of adjustable rotation parameters to achieve equivalent expressive transformation.
WiMi uses parameterized quantum circuits (PQC) and builds a deeply controllable quantum state transformer through layer stacking. The quantum circuit includes entanglement structures to ensure sufficient information flow between qubits, forming higher-dimensional representation capabilities than classical linear transformations.
The third step is decoding the quantum state back into a classical tensor. The results obtained from quantum measurement are reconstructed through a classical integration and correction module and finally returned to the decoding path of the classical U-Net. The features compressed through the quantum bottleneck retain expressive power yet complete the filtering and abstraction of high-dimensional information with an extremely low number of parameters. The entire process can be directly embedded into existing models without modifying the U-Net architecture or changing the training paradigm, achieving true “plug-and-play quantum enhancement”.
The release of WiMi's QB-Net marks a key step forward for our company on the path of quantum AI technology. It not only proves that quantum computing can deliver real value right now but also demonstrates the enormous potential of deep integration between quantum technology and deep learning. In the future, hybrid quantum-classical architectures will no longer be regarded as transitional technologies but will become one of the mainstream forms of AI for a long time to come.
QB-Net represents a brand-new way of thinking: letting quantum computing become the most valuable part of artificial intelligence rather than the entirety. The hybrid deep learning framework based on parameter-efficient quantum modules will bring a new structural optimization paradigm to the global AI industry and provide a completely new performance improvement path for enterprise-level intelligent systems.

About WiMi Hologram Cloud

WiMi Hologram Cloud Inc. (NASDAQ: WiMi) focuses on holographic cloud services, primarily concentrating on professional fields such as in-vehicle AR holographic HUD, 3D holographic pulse LiDAR, head-mounted light field holographic devices, holographic semiconductors, holographic cloud software, holographic car navigation, metaverse holographic AR/VR devices, and metaverse holographic cloud software. It covers multiple aspects of holographic AR technologies, including in-vehicle holographic AR technology, 3D holographic pulse LiDAR technology, holographic vision semiconductor technology, holographic software development, holographic AR virtual advertising technology, holographic AR virtual entertainment technology, holographic ARSDK payment, interactive holographic virtual communication, metaverse holographic AR technology, and metaverse virtual cloud services. WiMi is a comprehensive holographic cloud technology solution provider. For more information, please visit http://ir.wimiar.com.

Translation Disclaimer

The original version of this announcement is the officially authorized and only legally binding version. If there are any inconsistencies or differences in meaning between the Chinese translation and the original version, the original version shall prevail. WiMi Hologram Cloud Inc. and related institutions and individuals make no guarantees regarding the translated version and assume no responsibility for any direct or indirect losses caused by translation inaccuracies.

Investor Inquiries, please contact:

WIMI Hologram Cloud Inc.
Email: pr@wimiar.com

ICR, LLC
Robin Yang
Tel: +1 (646) 975-9495
Email: wimi@icrinc.com

FAQ

What did WiMi announce on January 2, 2026 about QB-Net (WIMI)?

WiMi announced QB-Net, a hybrid quantum-classical U-Net embedding a Quantum Bottleneck Module that reportedly cuts bottleneck parameters by up to 30x while keeping comparable performance.

How does the QB-Net Quantum Bottleneck Module work in WIMI's design?

It encodes classical features into quantum states, transforms them with parameterized quantum circuits including entanglement, then decodes measurements back into classical tensors.

Will QB-Net require changes to existing U-Net models for WIMI integrations?

According to the announcement, QB-Net is designed to be plug-and-play and can be embedded without modifying U-Net architecture or training paradigms.

What limitation did WiMi note about quantum AI for WIMI shareholders?

WiMi noted that current quantum hardware still cannot support large-scale quantum neural networks, so QB-Net uses module-level quantum enhancement rather than full quantum models.

How many parameters does a QB-Net quantum circuit use compared with classical bottlenecks?

The company states classical bottlenecks can have hundreds of thousands to millions of parameters, whereas the quantum circuit uses tens to hundreds of adjustable rotation parameters.