These machines do not provide any computational speedup, we have plenty of papers coming from academic physicists clearly showing this. Why these people insist with their narrative is beyond my comprehension at this stage (maybe, artificially increase the value of their market shares?)..
P.S. I was involved in semiconductor physics and quantum computing way before it became a buzz word.
Yes but if you’re involved in physics you know that physics papers are rich in discourse with correct and incorrect ideas alike.
With Phoenix IBM has a system with over 100 qubits close to 99.9 2Q Fidelity and though that does not sound like much it’s already enough to run operations that can’t be computed classically. The system is so well built that most of these qubits can perform well without crosstalk errors hindering execution, and with phoenix they solved mid circuit reset which didn’t work well before, blocking experiments for meaningful runtime error correction.
I also have to add that, so far, we know only a handful of algorithms which might run on a general quantum computer (if we had one!). Among those, only two performs something which would actually be useful (Grover and Shor). And we have only those, not for a lack of trying.. it's because coming up with quantum algorithms is intrinsically complicated (it's definitely not like writing an algorithm for a classical machine). So those quantum bits could be, in theory, the most stable thing in the universe, I'd still see them as irrelevant in everyday applications.
I agree with the “in everyday applications” qualifier. But there are plenty of applications that are research focused where having access to a proper quantum computer would be useful. As a Quantum Circuit fundamentally models a Unitary transformation on some quantum state it is hugely useful to model the time evolution of quantum systems which are experimentally difficult to access using a quantum computer.
Because peeps are doing HNDL (harvest now, decrypt later) today. They are intercepting your traffic and storing it until they can break crypto. They're doing that today. And they've been doing it for a while.
There are contexts in which that's an unacceptable level of risk.
HNDL can be worth it without the "threat" of quantum computing; for instance we see currently a lot of interest in AI-assisted math, and maybe it could lead to a breakthrough that affect crypto algorithms that are currently considered safe.
Therefore it seems weird to use QC as a benchmark. Rather, if your communications being decrypted in the future is unacceptable, then the logical conclusion is that you must not transmit it in such a way that a third party can intercept it.
Admittedly, it is inconvenient or even impractical, so people do accept the risk anyway without admitting it. They are sweeping the issue under the carpet of supposedly QC-proof crypto - because new quantum algorithms can be discovered too.
I'm personally more worried about flaws in the quantum-safe crypto schemes than in RSA, which is based on the age-old factorization problem of prime numbers.
They can harvest all they want, but if no one ever invents a usable, practical and scalable quantum processor all this will be moot.
What are the odds that a new QS crypto scheme has a fatal flaw or is otherwise intentionally backdoored?
Amping everyone up with fears over quantum breaking classical seems like a fantastic way to get another round of vulnerabilities injected into everyone's infrastructure.
I don’t understand who the target audience of these ads is. For most people, including myself, this is like seeing an ad for “New and Improved Time Machine! (Untested)” or “Magic Beans - Indica”.
These ads are aimed at the board members of IBM's customers. IBM has been doing this for ages. I might think IBM is a high margin body shop with a veneer of innovative-ish projects like Watson that are almost comically out of step with the rest of the tech industry. But that's not the golf course discourse.
The purpose is to make it safe to remain an IBM customer.
Everyone who is curious? I mean did you know that you can run something on a real quantum computer for free? Go try out something new, or are we only getting excited about the newest AI slop machine release these days?
They are currently targeting more material science, such as quantum chemistry for drugs discovery or new material discovery. They can’t and do not aim to do prime factorization on those.
Quantum software engineer here - yes! I use Claude quite extensively with quantum SDKs like Qiskit and Pennylane. It tends to be very good at scaffolding and architecture, but can spiral or be outright incorrect when it comes to implementing certain algorithms like QAOA.
The real benefit comes from real-world integration - if I'm exploring e.g. a logistics application, it can source real-world data and models very quickly, and suggest quantum appropriate-formulations. Nevertheless, a lot of hand-holding is required.
P.S. I was involved in semiconductor physics and quantum computing way before it became a buzz word.
With Phoenix IBM has a system with over 100 qubits close to 99.9 2Q Fidelity and though that does not sound like much it’s already enough to run operations that can’t be computed classically. The system is so well built that most of these qubits can perform well without crosstalk errors hindering execution, and with phoenix they solved mid circuit reset which didn’t work well before, blocking experiments for meaningful runtime error correction.
There’s an open scoreboard ibm runs open to all to submit problems demonstrating algorithmic breakthroughs https://quantum-advantage-tracker.github.io/
There are contexts in which that's an unacceptable level of risk.
Therefore it seems weird to use QC as a benchmark. Rather, if your communications being decrypted in the future is unacceptable, then the logical conclusion is that you must not transmit it in such a way that a third party can intercept it.
Admittedly, it is inconvenient or even impractical, so people do accept the risk anyway without admitting it. They are sweeping the issue under the carpet of supposedly QC-proof crypto - because new quantum algorithms can be discovered too.
They can harvest all they want, but if no one ever invents a usable, practical and scalable quantum processor all this will be moot.
Amping everyone up with fears over quantum breaking classical seems like a fantastic way to get another round of vulnerabilities injected into everyone's infrastructure.
The purpose is to make it safe to remain an IBM customer.
Here they mix both and don’t get the public. It has always been IBM problem to correctly target the product to the audience.
We see it here again
https://quantum.cloud.ibm.com/computers?system=ibm_phoenix
Does this mean they literally have a single one? (Or at least a single one they offer publicly)
https://towardsdatascience.com/where-are-we-with-shors-algor...
If you run it over and over again you will eventually get 3 or 5...
The real benefit comes from real-world integration - if I'm exploring e.g. a logistics application, it can source real-world data and models very quickly, and suggest quantum appropriate-formulations. Nevertheless, a lot of hand-holding is required.
Also from IBM: "Bob"