Photo PsiQuantum
Until now, we have purposefully remained quiet about our approach and progress. In the interest of advancing the quantum computing conversation, we thought we could use this moment as an opportunity to share an overview of our thesis, our progress and our path forward.
Building a fully programmable quantum computer is one of the most exciting current challenges in science and technology. In the future, quantum supercomputers will be able to model chemical reactions and the properties of materials with a level of accuracy that is, and will remain, forever impossible with conventional computer technology. That has the potential to transform every corner of modern technology: the cars we drive, the planes we fly in, the medicines we take, the industrial processes that produce everything from fertilizers to batteries for electric vehicles. The most promising applications are probably still to be discovered.
Our approach to quantum computing: you need a million qubits
For a quantum computer to be useful it has to be able to carry out complex calculations that require many billions of “gate operations”. Quantum states are extremely fragile, which is the reason the quantum processes happening in the world around us are invisible. If we do not correct for errors then with so many gate operations, our computer would output data ruined by random noise.
A lot of error correction is required to protect qubits, and we believe the first useful quantum computer requires at least 1 million physical qubits. Even a machine of one million qubits will comprise many billions of components, similar to today’s classical computers. To ensure that we can deliver such a complex device our architecture is unique, using only manufacturing processes that can be implemented in the same tier 1 semiconductor fabrication facilities that produce your laptop and cellphone. Recent advances in semiconductor manufacturing and silicon photonics have unlocked new methods to build qubits in this way.
Other approaches to quantum computing are matter-based. Prominent examples use superconducting devices or ion traps to produce qubits, rather than light. These approaches are beginning to enable Noisy Intermediate Scale Quantum (NISQ) devices with more than 50 qubits — but without large-scale error correction or a clear path to manufacturing million-qubit devices. The exciting applications for quantum computers mentioned above cannot be implemented with NISQ devices.
Rather than take a quantum system and try to make it scalable, we have taken a scalable process — silicon manufacturing — and made it quantum.
A photonic approach to quantum computing has many advantages in addition to manufacturability. Photonic qubits are inherently low noise and do not interact in uncontrolled ways. They can operate at higher temperatures which means we can place more control electronics on a quantum chip, improving system integration. Modularity and networking are critical elements of scalable error correction. We achieve both, through the manufacturability of our system, and because we can readily send qubits between chips using conventional optical fiber. Matter-based approaches need complicated devices to convert qubits into photons to connect modules. Our approach cuts out the middleman — our qubits are always photons.
Our progress and our path forward
The path to a useful quantum computer will be a long but worthy one, necessitating the considerable development of system architecture, numerical modeling, hardware integration, control systems and operating software.
We have assembled a team of more than 100 engineers with expertise across all aspects of silicon manufacturing and error corrected quantum computing. We are also proud partners with a Tier 1 semiconductor production fab, Global Foundries, which has accelerated our progress in the last year.
To deliver 1 million physical qubits quantum computers need to be large-scale, error-corrected and manufacturable. These principles are critical to success, and our belief is that the path forward is photonics.
Building a fully programmable quantum computer is one of the most exciting current challenges in science and technology. In the future, quantum supercomputers will be able to model chemical reactions and the properties of materials with a level of accuracy that is, and will remain, forever impossible with conventional computer technology. That has the potential to transform every corner of modern technology: the cars we drive, the planes we fly in, the medicines we take, the industrial processes that produce everything from fertilizers to batteries for electric vehicles. The most promising applications are probably still to be discovered.
Our approach to quantum computing: you need a million qubits
For a quantum computer to be useful it has to be able to carry out complex calculations that require many billions of “gate operations”. Quantum states are extremely fragile, which is the reason the quantum processes happening in the world around us are invisible. If we do not correct for errors then with so many gate operations, our computer would output data ruined by random noise.
A lot of error correction is required to protect qubits, and we believe the first useful quantum computer requires at least 1 million physical qubits. Even a machine of one million qubits will comprise many billions of components, similar to today’s classical computers. To ensure that we can deliver such a complex device our architecture is unique, using only manufacturing processes that can be implemented in the same tier 1 semiconductor fabrication facilities that produce your laptop and cellphone. Recent advances in semiconductor manufacturing and silicon photonics have unlocked new methods to build qubits in this way.
Other approaches to quantum computing are matter-based. Prominent examples use superconducting devices or ion traps to produce qubits, rather than light. These approaches are beginning to enable Noisy Intermediate Scale Quantum (NISQ) devices with more than 50 qubits — but without large-scale error correction or a clear path to manufacturing million-qubit devices. The exciting applications for quantum computers mentioned above cannot be implemented with NISQ devices.
Rather than take a quantum system and try to make it scalable, we have taken a scalable process — silicon manufacturing — and made it quantum.
A photonic approach to quantum computing has many advantages in addition to manufacturability. Photonic qubits are inherently low noise and do not interact in uncontrolled ways. They can operate at higher temperatures which means we can place more control electronics on a quantum chip, improving system integration. Modularity and networking are critical elements of scalable error correction. We achieve both, through the manufacturability of our system, and because we can readily send qubits between chips using conventional optical fiber. Matter-based approaches need complicated devices to convert qubits into photons to connect modules. Our approach cuts out the middleman — our qubits are always photons.
Our progress and our path forward
The path to a useful quantum computer will be a long but worthy one, necessitating the considerable development of system architecture, numerical modeling, hardware integration, control systems and operating software.
We have assembled a team of more than 100 engineers with expertise across all aspects of silicon manufacturing and error corrected quantum computing. We are also proud partners with a Tier 1 semiconductor production fab, Global Foundries, which has accelerated our progress in the last year.
To deliver 1 million physical qubits quantum computers need to be large-scale, error-corrected and manufacturable. These principles are critical to success, and our belief is that the path forward is photonics.