How we build this list
Every company here is placed by what it ships: a physical qubit modality, an SDK, a cloud access layer, or a specific application, not by marketing category. Public companies are tagged with their ticker. Well-funded private companies and university/government programs are included when they have a public roadmap or a way for outsiders to use their technology.
We don't take payment for placement or ranking, and this page isn't a leaderboard. A startup and IBM sitting in the same grid says nothing about scale. If you want to run something today instead of reading about the industry, start with 模拟器 or 真实硬件 instead.
超导硬件
使用冷却至接近绝对零度的超导 transmon 量子比特来构建量子处理器的公司。
全球最大的量子计算机队——IBM Quantum Network 上有 100+ 台 QPU。包括 Eagle(127 量子比特)、Osprey(433 量子比特)、Condor(1121 量子比特)、Heron。提供免费套餐。
Sycamore 处理器实现了量子霸权(2019 年)。Willow(105 量子比特)展示了低于阈值的纠错能力(2024 年)。研究访问需申请。
超导量子云的先驱。Aspen-M 系列多芯片 QPU。可通过 Rigetti QCS 和 Amazon Braket 访问。
Azure Quantum 云平台 + 拓扑量子比特研究(Majorana 1 芯片,2025 年)。托管 IonQ、Quantinuum、Rigetti 硬件。
离子阱硬件
利用电磁场捕获单个离子的量子计算机——通常拥有目前可用的最高门保真度。
由 Honeywell Quantum Solutions 分拆而来。H 系列离子阱系统保持多项第一名量子体积记录。TKET SDK 开源。
光子硬件
使用光子(光的粒子)作为量子比特的量子计算机——可在室温下运行。
光子量子计算 + PennyLane SDK。Borealis(216 个压缩模式)展示了量子优势(2022 年)。Strawberry Fields 框架。
中性原子硬件
使用激光冷却的中性原子并将其保持在光镊中的量子计算机——具有高连接度和长相干时间。
哈佛/MIT 分拆企业。Aquila(256 个中性原子量子比特)已上线 Amazon Braket。展示了 48 个逻辑量子比特(2023 年)。
量子软件与 SDK
用于构建和运行量子电路的开源 SDK、编译器和量子编程框架。
使用最广泛的量子 SDK。50 万+ 注册开发者。Apache 2.0 开源。Qiskit Runtime 用于真实硬件。
面向 QML 的可微分量子计算。支持 PyTorch、JAX、TensorFlow。30+ 设备后端。
抽象 Qiskit、Cirq、PennyLane、Braket、CUDA-Q、IonQ 的统一 API。一次编写,可在任意后端运行。
云平台
为多个量子硬件提供商和模拟器提供统一访问入口的云服务。
最大的量子云计算机队。Open Plan(免费,127 量子比特 Eagle)+ Premium 套餐。IBM Quantum Learning 平台。
微软量子云。IonQ、Quantinuum、Rigetti 硬件。新账户可获 Azure Quantum Credits。
通过 Google Cloud 访问 Google 的 Sycamore 和 Willow 处理器。需申请研究项目。
量子算法与应用
专注于量子算法、优化以及特定行业量子应用的软件公司。
纠错与控制
专注于量子纠错、硬件控制电子设备以及容错量子计算的公司。
量子密钥分发与安全
构建量子密码硬件、QKD 系统和后量子安全产品的公司。
量子传感与计量
将量子力学应用于超精密传感、导航、授时和成像的公司。
量子计算(退火)
针对组合优化问题进行优化的量子退火处理器。
Quantum Computing Inc. 及其他上市公司
其他在量子计算领域占据重要地位的上市公司。
国防与航空航天
拥有活跃量子计算项目的国防承包商和航空航天公司。
金融与银行
拥有专门量子计算研究团队和生产试点项目的金融机构。
学术与研究项目
运营免费量子计算访问项目的大学和研究机构。
10 亿欧元的欧盟计划。资助 OpenSuperQ、AQTION、IQM、Pasqal 等众多项目。研究和工业实验室。
Reading the landscape: what's happening
No single hardware modality has won. Superconducting qubits (IBM, Google, Rigetti) currently lead on qubit count and software maturity, trapped-ion (IonQ, Quantinuum) leads on gate fidelity and connectivity, neutral atoms (QuEra, Pasqal) are scaling fast with a simpler cooling stack, and photonic approaches (Xanadu, PsiQuantum) bet on room-temperature operation and fiber-network compatibility. Each trade-off (coherence time, gate speed, connectivity, cooling cost) still favors a different modality, which is why hardware R&D funding stays spread across all four rather than consolidating.
Access has consolidated even though hardware hasn't. A few years ago, using a given QPU meant signing up with that vendor directly. Today, IBM Quantum, AWS Braket, Azure Quantum, and Google Quantum AI act as gateways that expose multiple hardware backends (often including competitors' machines) through one account and one SDK. That's the main reason a free-tier guide like this site's 真实硬件 page can cover most of the industry without listing dozens of separate signups.
The software and algorithms layer is shifting from general-purpose demos toward vertical specialization: chemistry simulation, portfolio optimization, and logistics/routing are where most "quantum algorithms & applications" companies now focus, rather than claiming broad quantum advantage across every industry at once.
Error correction and post-quantum cryptography are the two fastest-growing categories on this page for the same underlying reason: raw physical qubit counts stopped being the bottleneck story, and both fault-tolerant computing and the security response to it depend on solving error rates, not adding more noisy qubits.