QPU at home?
Global Quantum Intelligence, LLC
The leading Quantum Tech market & business intelligence provider. Advising leaders, investors and governments globally.
Quantum Computing Shipments for On-Premise Installation are Starting to Increase and That Will Add New Requirements for Processor Design
One thing we have started to notice over the past year is a few companies are starting to talk about on-premise quantum computer shipments. The number is still quite small, but this may be a start of a trend we will see more often in the coming years. Just a couple of years ago, the feeling was that almost all quantum computer access would be through the cloud because of the difficulties of supporting and maintaining a remote quantum computer. These machines are finicky and require a lot of manual calibrations, hand holding, and close availability of the engineers who built them to keep them running. Several years ago, D-Wave had shipped machines for customer use by end customers to the University of Southern California (USC), Google/NASA, Los Alamos National Laboratory, and Temporal Defense Systems, and then stopped doing that for a while. But they did have one installation last year at the?Forschungszentrum Jülich Supercomputing Centre?in Germany to provide better cloud support for their European customers.
However, there are some customers who prefer having a machine on-premise rather than just using one in the cloud. There are a number of reasons for this including national pride, data sovereignty concerns, data security, desire to avoid waiting in a queue to run their jobs, and a need to perform a large volume of quantum processing that would justify having a dedicated machine for internal use. In the future, there may be some complex quantum algorithms that may take a several weeks to complete. Although this may seem like a long time, there could be a high value use case, drug discovery for example, that would be impossible to complete classically, but could result in a new billion dollar product, if the quantum approach can find the right solution.
Here is a partial list of recent announcements we are aware of that involve on-premise quantum processor shipments.
While most of the examples listed above will be installs used for research and training purposes, in the future we expect there will be more demand from end user organizations that will want to use the processors for real commercial work.?And, in some respects, these end user organizations will be more demanding.?They will want their machines to provide a high?MTBF (Mean Time Between Failures), 24/7 uptime, as little downtime as possible for calibration and maintenance, and quick response from the vendor for any problem that arises.
Classical computing companies have been used to these requirements for many years and design into their product features for?Reliability, Availability and Serviceability?(RAS). Quantum companies will need to learn how to do this too. One of the techniques they will need to adopt (and some already have) is to develop automated calibration and testing algorithms so that you don’t need a PhD to bring up a machine from a cold start. Another technique is to design the machine with?Field-Replaceable Unit?(FRU)?so that if something breaks a technician can quickly and easily make the repair. You don’t want to have a technician spend many hours disassembling a machine to get at some component that is hidden deep inside the processor.
领英推荐
Having an adequate stockpile of spare parts is another consideration. Some components used in quantum processors like lasers and dilution refrigerators may have 6-12 month or more lead-times. No end user would stand for that much delay in getting a machine repaired. Many classical computing systems have a capability for remote monitoring. Someone offsite can log in and find out the general health of the system to ensure it is operating properly. And sometimes, if something is wrong they can provide a software fix to get it resolved without an on-site visit. But on-site visits will never be totally eliminated and, at some point, a company may need to set up a team of Field Service Engineers who will visit the remote sites to make repairs.
Right now, may of the quantum processor companies are just concentrating on getting their system to work. Progress is currently measured in how well the design is doing with regards to the number of qubits, the quality of the qubits, and the speed of the qubits. But as the machines get more capable, quantum processor companies will need to incorporate these Reliability, Availability and Serviceability?(RAS) requirements starting during the early design stage of a new processor, if they want to pursue on-premise sales.
Retired Technologist, Freelance Consultant, and Software Developer
1 年What about ownership? The quantum computer system may be on-premise, but is it owned by the customer or leased from the vendor? And what about service life? Will the vendor take back the expired system for free, or at what cost? And if leased, what term of commitment, monthly/annual lease cost, and what cost to get out of the lease early? And is there a maintenance contract? At what cost? What about insurance? Will the vendor take care of that, or what might it cost the customer to insure the system? What would the insurance cover - flood, fire, earthquake, theft, physical damage (hit by a forklift, vandalized, or falling objects), and... theft?
Merging Data with Intuition ?? ?? | AI Assistant Professor @ Colegio de Matemáticas Bourbaki | Quantum Computing Ms. Sc. Thesis Student @ DUTh | EMBA Candidate @ Valar Institute
1 年The company which puts a QPU in every house as NVIDIA has done with GPU’s is the one who is going to take the market by storm. Please at room temperature ??