- 03/21/2024
There is a lot of hype that quantum computers will jeopardize our data security and privacy. Is there truth to the hype? Will quantum computers break current encryption standards?
Elton: In time, as quantum computers mature, they will more than likely be able to break encryption standards that we rely on today for everything from banking to health to cryptocurrency.
Vlad: When people talk about security aspects of quantum computing, that area encompasses both challenges and opportunities.
Typically, when people talk about quantum security, it's about breaking cryptography and what needs to be done in order to mitigate this. When we talk about cryptography, there are many misconceptions. Quantum cryptography won't be able to break everything. It doesn't break symmetric cryptography – cryptographic protocols that rely on each party having their own copy of a secret key. It may require us to use longer keys, but it doesn't fundamentally break it. It also doesn't break hash codes used to build blockchains and it's the wrong idea that once we have a quantum computer it will be able to reveal the hash sequence of hash code and rebuild the blockchain. However, it will likely compromise signatures on the blockchain, as these use an asymmetric cryptographic mechanism, which is when the protocol involves both a public and private key.
Asymmetric cryptography is ubiquitous. For example, when you go to almost any website, you have a certificate published in the address line, and you can open it to see two main pieces: the public key and the signature that was generated on behalf of the owner of the public key. If the public key protocol is broken, this would allow malicious actors to decrypt information that’s meant to be securely transmitted between you and the certificate owner.
All asymmetric encryption methods are built upon so-called “hard problems”, mathematical problems that would require thousands or even millions of years to break with classical computers. Ultimately quantum computers can solve some of these problems in a much more direct fashion, making the encryption breakable in days or even minutes instead of millions of years.
Now, even though there are no practically used quantum computers right now, some information and data may still be at risk if it has long term value. For example, financial, medical or reputational information or cryptocurrency keys may still be useful in 20 years. And that means when those pieces of data are being exchanged today, they can be intercepted. Even though in encrypted form, they can be stored and then decrypted later when quantum computers are available. This is called a “harvest now, decrypt later” threat.
There is a lot of hype that quantum computers will jeopardize our data security and privacy. Is there truth to the hype? Will quantum computers break current encryption standards?
Elton: In time, as quantum computers mature, they will more than likely be able to break encryption standards that we rely on today for everything from banking to health to cryptocurrency.
How do we mitigate that kind of risk?
Vlad: To mitigate those risks imposed by quantum computing, there are two major approaches. One is called Post-Quantum Cryptography or Quantum Resilient Cryptography, and another one is Quantum Key Distribution. We, in FCAT, analyze what already exists in these areas because there are different approaches, techniques and equipment. Then, we determine how we can integrate pieces of technologies in the overall ecosystem within Fidelity. For example, we’re creating APIs that can be compatible and that can be integrated with existing systems like identification and access management systems. We also work with hardware equipment to evaluate what's available, as well as pros and cons, and gaps in existing technology. Last, but not least, we work on technology adoption by planning and building roadmaps of how to migrate from existing practices in security to new quantum-resilient practices.
Do you find the relationship between quantum computing and generative AI to be adversarial?
Mike: Generative AI has seen a lot of pressure this year, and I think that has impacted, perhaps negatively, a lot of the messaging and some of the research coming out of the quantum ecosystem. Though, long term, these aren't exclusive - these are two technologies that are probably going to work together in the long term. There might be some adversarial nature in funding but I’m hopeful it will be short term. But yes, there's no doubt that the hype around AI has impacted the quantum ecosystem.
Elton: I think a lot of things have an adversarial effect on each other. There's nothing in particular that is adversarial between quantum and generative AI. These are all amazing technologies that try to help mankind and advance our civilization.
What does the future look like with quantum and AI working together?
Elton: Quantum computers may help to accelerate components of AI or create more powerful models, but this will require a more mature quantum computer, which is not available yet. We can’t expect to say, ‘I want to put my AI system on a quantum computer’ and just expect that to work better or faster, that's not how it works.
A lot of AI systems take millions of dollars to train, and probably a couple of months just to train a single iteration. So, we imagine that if large scale quantum computers become available, it's possible that can make it easier to train AI systems. How much faster, whether that's goes down from months to days or days to hours, that we don't know yet. It highly depends on advances in the hardware. It’s possible that with quantum computers embedded into a supercomputer, it could create more powerful AI models, but whether that is desired is another question. Do we need a better approach? There's a more ethical question there.
How do you see quantum computing being powered in terms of energy?
Elton: There are a few different modalities of quantum computers. There’s what is called a superconducting quantum computer, which needs a dilution fridge to bring down the temperature of the quantum chip to near absolute zero to make it work. That kind of quantum computer will require a lot of power - think megahertz of power. Whether that's the way forward or not, we don't know. There are other modalities of quantum computers that claim that in 2024 and 2025 they can put rack-mountable quantum computers onto data centers. That means the power consumption of their quantum computers would be a lot lower. Ultimately, some will require more power than others. Power consumption is definitely top of mind for a lot of people.
How do you determine what problems quantum computers are best suited for?
Elton: There are standard ways to identify if an application or if a business problem will benefit from quantum computers. Typically, we look at problems in optimization, machine learning and Monte Carlo simulation because these are the areas that we think that quantum computers will be able to help.
I think the easiest way to tell is if you have anything you want to go faster. But we're not just looking at quantum, we’re looking at the whole spectrum of future computing. In fact, most of the time we see that problems don't benefit from quantum computers but there are other ways to massively accelerate such systems with classical hardware, for example, using state-of-the-art GPUs or distributed computing.
Vlad: I would add to that anywhere where you use random number generation, for example in generating cryptographic material, you can likely also benefit from quantum devices.
What would it take to get to quantum computer maturation and adoption sooner?
Elton: On the national level, support for start-ups and labs that aim to commercialize the technology would be helpful. For enterprise users, I think the best thing we can do is to discover more use cases that will benefit from such technologies so that we can continue to justify research and development in this area.
Vlad: I would add that we have several use cases which can be implemented right now or very soon in the security space. It would be good if we started exploring them in practical implementation, not only in the lab but in production or pre-production environments.
There are always legal considerations and adoption beyond a national level. What are you seeing on an international level when it comes to quantum computing?
Mike: There are very different approaches to quantum security across the globe. The American government’s message has focused entirely on post quantum cryptography, and generally, they don't recommend QKD at all.
This is very different from most other governments, companies, and researchers, where efforts are much more hybridized (combining post-quantum cryptography and quantum key distribution). For example, Europe has established a very strong investment in creating a quantum network across the continent with all 27 member states having joined in. Similarly, we know Singapore, South Korea, and China are also using QKD, sometimes even reporting satellite-enabled QKD experiments. The UK is also using it. So, most of the big programs outside of the US seem to be hybrid.
How vulnerable are satellites to being hacked?
Vlad: There is no perfect security. Satellites by themselves can definitely be hacked. But if we look to the next generation of quantum key distribution, which will use a quantum phenomenon called entanglement, you don't really have to trust the satellites (or other nodes) in your network. The nice thing about entanglement is that it allows so-called device independent distribution without trust.
The problem is it's mostly at the lab level right now. Even though you can hack into it, the important aspect is you as a user can detect that someone is actually tampering with the device and that allows you to act appropriately. With classical cryptography, you don't know. Somebody can tap on the wire with data, maybe translate them, maybe store them somewhere and you have no idea.
Are there opportunities to collaborate with the Fidelity Center for Applied Technology on quantum or future computing research and development?
Mike: Yes! If you’re in Fidelity and have a computationally difficult problem, we want to hear about it. In the past we’ve successfully helped teams speed up processes to save time, money, and effort - sometimes even turning slow prototypes to full production-ready tools. And if you’re outside Fidelity and have developed a new technology in this space, please reach out to us at FCAT@fmr.com. We’re always eager to look at benchmarked advances to see whether they may prove advantageous.
