AWS VP of engineering on the cloud giant's first major quantum computing push
Amazon Web Services (AWS) has used the first day of its annual Re:Invent conference in Las Vegas to set out its plans to democratise quantum computing, while also firming up the enterprise use cases for the technology
Amazon Web Services (AWS) has thrown its hat into the quantum computing ring, after four years in stealth mode, as it has fought to bring to market a proposition it hopes will cut through the hype surrounding the technology.
That’s according to the cloud giant’s vice-president of engineering, Bill Vass, who sat down with Computer Weekly on the first day of the annual AWS Re:Invent conference in Las Vegas to talk through the firm’s first major push into the field of quantum computing.
It is an area a handful of its cloud-touting contemporaries are already playing in, including IBM, Microsoft and Google, and much of the work being done here is still exploratory and experimental in nature, focused on testing the limits of what the technology can do.
As is often the case with any technology area that the major players are rushing to invest in, there is a lot of hype surrounding quantum computing, and Vass said the company is treading carefully to avoid adding to it unnecessarily.
“I’m pretty bullish on the long-term capabilities of these machines. In the short term though, it’s been overhyped,” he said.
“We’re trying give people a practical interaction with it, and will continue to focus on areas such as a Hamiltonian simulation for a molecule, or things that we know quantum computers will do well.”
Out of stealth and into quantum computing
For AWS, its move into the quantum computing space is four years in the making, said Vass, and without much fanfare either.
“We haven’t spoken much about it because it’s pretty early days in quantum computing, but our real intense efforts have been over the past year or so, as we’ve been investing and working with partners and figuring out the best way that we can do things practically with customers,” he said.
The end result is the preview release of Amazon Braket, which was announced on the first day of Re:Invent. This is a managed service geared towards enabling scientists, researchers and developers to access cloud-based quantum computing resources where they can test their own qubits and circuits.
According to Vass, the setup marks a continuation of AWS’s ongoing commitment to lowering the barriers to entry for technologies that have traditionally – from a cost, complexity or skills perspective – proven too high for organisations to adopt, despite the benefits they stand to gain by doing so.
For example, the cloud-based quantum computing resources Amazon Braket provides access to are run off hardware provided by the likes of D-Wave, IonQ and Rigetti. This hardware is costly to deploy, and needs to be built and operated in specialised, temperature or pressure-controlled environments the average enterprise would struggle to build themselves.
“Users don’t have to then go buy a machine or set up, or evaluate iterations or set up a partnership or any of those kinds of things. They just select to use it and it shows up on their AWS bill just like any other compute resource,” he said. “It just looks like any other component in the cloud to them. And that’s part of our goal.”
Users can sign up to use the preview version of Amazon Braket now. From 16 December 2019, the firm will start running jobs with interested parties “on a first-come, first-served” basis, said Vass, before the service enters general availability during the first half of 2020.
“With these early machines, you’ve got about 32 qubits available and you’ll be able to – on the annealing side – do some good, optimisation algorithms,” he said.
“On the gate-based machine side, you won’t get results you couldn’t also get on an analogue computer with a good set of mathematics, but now is a great time to start learning and experimenting and see how it would integrate into your own IT environment.”
In this regard, it is AWS’s aim to make the integration between users’ IT environments and the quantum computing-based ones as seamless as possible, he added.
“Similar to how a FPGA [field programmable gate array] would work, where you have the standard digital processing that you’re doing and then when you want something that’s accelerated to do factoring very well or something that a quantum computer does really well, so you make a call to this machine that operates your shot on it and it gives you the results back,” he said.
The roll-out of Amazon Braket is also underpinned with professional services support in the form of Amazon Quantum Solutions Lab. This is an initiative that will see users partnered up with the firm’s quantum computing specialists, or those working with its consulting partners, to help organisations identify use cases for the technology.
“There’s a whole bunch of things that are possible, if we understand the low-level molecular reactions much better. And those are things that are very hard to do on a high-performance computer that a quantum computer does very well,” said Vass.
“The biggest areas where the hard problems are [for customers] is around Hamiltonian Simulations, and a quantum computer is particularly good at those because that’s the simulation of how a molecule works because a molecule works using quantum mechanics.”
We’ve been looking at how we would use [logistics optimisation] internally to optimise our environments
Bill Vass, AWS
For organisations operating in pharmaceutical or material sciences space, this could lead to advances in drug development, as well as in the oil and gas sectors, and manufacturing too.
“For example, if you can create a full end-to-end simulation of ammonia, you could save potentially billions in factoring costs, and that could be [hugely valuable].”
Within the walls of Amazon, quantum computing has been pinpointed as means of optimising the logistics side of its parent company’s online retail operations, said Vass.
“Logistics optimisation is a big deal for us, and some of these quantum algorithms can really help there so we’ve been looking at how we would use that internally to optimise our environments,” he said.
This is a pattern of development is a regular occurrence within Amazon, said Vass, who made the point that AWS was originally created to fulfil the cloud compute and storage needs of its retail arm, before being opened up to the public to use.
Another example of this is Amazon Kinesis Video, the firm’s real-time data streaming service that is designed to stream video from connected devices to AWS cloud where it can be processed using data analytics.
“Kinesis video was developed because of Amazon Go stores… and can process a million concurrent images or threads in real-time. It’s a very high-performance system, and it enables all the Amazon Go stores. A lot of people don’t realise anyone could build an Amazon Go using Kinesis Video today,” he added.
As Vass previously alluded to, it is still early days for quantum computing and there will be some features with regard to how Amazon Braket works, specifically, that it might take some long-standing users of AWS, accustomed to always-on tech a little bit of getting used to.
The machines underpinning the service, for example, will go down for half an hour every four hours so they can be recalibrated, although the development and simulation environments AWS will provide as part of the service will remain accessible 24 hours a day.
“You can do your development, your simulation, and all your testing and then when you’re ready to do your shots, they’ll all be queued up to run on the machines,” he said.
“Generally the shots run so fast… you can queue up [shots] during their calibration, and as the machines mature that will get better [and require] less calibration, and error corrections and they’ll be more qubits becoming available, and exposing them on the cloud will really help to improve their maturity.”
Read more about quantum computing
Volkswagen Group recently demonstrated how it was working with D-Wave to research uses for quantum computing. We find out what the car maker is hoping to achieve.
Industry experts predict it will take 10 years for quantum computing to become a reality, but Microsoft believes it has the research edge, with systems, software and technology to get there in five.
Read more on Clustering for high availability and HPC
