Haiqu Inc.
Haiqu Inc. is an American quantum software company headquartered in San Francisco, California, with a significant R&D presence in Ukraine. Founded in 2022, the company operates at the critical intersection of quantum hardware and practical applications.
The company's core mission is to accelerate the timeline to practical quantum computing. It directly addresses the primary obstacle holding back the industry: the "clumsy" nature of current quantum hardware. Today's quantum processors, often referred to as Noisy Intermediate-Scale Quantum (NISQ) devices, suffer from a limited number of qubits and extreme sensitivity to environmental "noise," which leads to high error rates and prevents them from running complex, valuable algorithms.
Continue…Haiqu is not a hardware manufacturer. Instead, it develops a sophisticated, platform-agnostic software layer (middleware) that acts as an intelligent execution engine. This software boosts the performance of existing quantum computers, allowing them to run algorithms that are significantly deeper and more complex than would otherwise be possible.
Core Focus: Overcoming Hardware Limitations
Haiqu's entire business model is built on solving the noise and scale problem of near-term quantum processors. The company's technology is designed to "extract value out of clumsy near-term quantum hardware" by enabling quantum applications that were previously considered impossible.
The company's approach is twofold:
1. Enhance Hardware Capability: Develop software that intelligently breaks down, optimizes, and executes quantum circuits to mitigate the effects of noise and hardware limitations.
2. Enable High-Value Applications: Focus on use-cases, particularly in Quantum Machine Learning (QML), where even today's noisy devices can be used to process high-dimensional data that is intractable for classical computers.
Products and Services
Haiqu's products are software-based solutions for quantum computers, aimed at both hardware providers and application developers.
Quantum Application Execution Stack
This is Haiqu's flagship product, a software layer that manages the execution of quantum algorithms. It is designed to be hardware-agnostic, meaning it can work with various types of quantum processors. This stack includes several proprietary technologies:
- Approximate Circuit Execution: A "divide-and-conquer" methodology that allows for the execution of quantum circuits that are dramatically more complex ("100x deeper") than the hardware would normally allow. It intelligently splits a large, deep circuit into multiple smaller, modified circuits that can fit within the constraints of the quantum processor. The results from these sub-circuits are then combined using advanced error mitigation to produce a high-fidelity final answer. This technique is particularly valuable for applications like computational fluid dynamics (CFD).
- Error Mitigation: The platform includes a high-performance, lightweight error mitigation system. This system is efficient enough to be used during variational routines (a common iterative quantum algorithm), which reduces the number of "circuit runs" needed to find a solution and improves the final fidelity of the result by 2-3 times.
Quantum Machine Learning (QML) and Data Encoding
This is a key application and service offered by Haiqu. The company has developed a breakthrough proprietary data encoding technique (also known as quantum feature embedding).
- High-Dimensional Data Encoding: The primary challenge in QML is loading large, complex classical datasets onto quantum computers with limited qubits. Haiqu's technology provides a novel method to compactly represent high-dimensional data (e.g., a financial dataset with over 500 features) onto a processor with as few as 128 qubits.
- Application in Anomaly Detection: Haiqu has famously demonstrated this capability on one of IBM's Quantum Heron processors. By using their quantum preprocessing technique on a high-dimensional financial dataset, they were able to improve the accuracy of a classical machine learning model in detecting fraud and anomalies. The quantum process was able to identify complex patterns and correlations in the data that classical methods missed.
- Beta Program: Haiqu is now offering beta access to this quantum feature embedding technology for companies in various sectors to explore its use in finance, healthcare (medical readings), industrial monitoring (faulty sensors), and geophysics (seismic data).
RIVET (Modular Transpilation SDK)
RIVET is an open-source software development kit (SDK) provided by Haiqu. Its purpose is to solve a key bottleneck in quantum workflows: transpilation. Transpilation is the process of translating a high-level quantum algorithm into the specific, low-level gate operations required by a particular quantum computer.
- Speeding up Workloads: For QML and other algorithms that require running a high volume of different circuits, this transpilation step can take hours. RIVET is designed to slash this time from hours to seconds, making the entire workflow dramatically more efficient.
