Amazon CTO talks about quantum computing: still in the early stage of development, focusing on reducing costs


Source: AWS
Reporter Peng Xin
A new force suddenly rises. Novel coronavirus pneumonia continues to hit. Video conferencing, online shopping and video games are all in the air. The whole world relies heavily on cloud computing supported digital infrastructure. In this context, e-commerce and cloud computing giant Amazon is growing rapidly.
During the recent AWS re: invent conference, Amazon CTO Werner Vogels gave an exclusive interview with interface journalists to look forward to the technology trends and explain the development ideas of AWS technology.
During the outbreak, many companies increased the use of cloud services, helping to increase the revenue of cloud service providers such as AWS and Microsoft, of which AWS accounted for the majority. Werner Weigel said that because of the impact of the new epidemic, it is difficult to achieve business in many physical ways, so we all use technology to achieve customer coverage, which is the rise of “software as a service”.
According to Gartner, a market research firm, AWS has a 45% share of the global public cloud market, more than the second to fifth places combined. According to the latest revenue data, AWS has ranked among the top five IT enterprises in the world with Microsoft, Dell, IBM and Cisco.
At present, advanced technology investment has become the focus of AWS in the field of cloud computing. It has made extensive deployment and investment, and launched a number of cloud services, with the purpose of increasing efforts to attract advanced technology customers, including government and institutional start-ups. AWS technology layout involves aerospace, quantum computing, high performance computing, etc. In July this year, AWS announced the establishment of an aerospace business unit. At the re: invent conference in December last year, AWS announced the launch of Amazon braket, a quantum computing service, and the establishment of AWS quantum computing center and Amazon quantum solutions laboratory.
For quantum computing, Werner Weigel stressed that AWS mainly uses the scale effect to reduce the cost of quantum computing as much as possible. In the case of quantum computing is still in a very early stage of development, the main players have limited willingness to compete.
On December 18, Werner Weigel delivered a speech on future technology trends and prospects, pointing out that in 2021, the advancement of cloud and machine learning to the edge will be further accelerated, embedded in portable devices, cars, homes and workplaces, and eventually the existence of cloud will be far-reaching. On the other hand, small businesses are competing in the cloud, and Southeast Asia and sub Saharan Africa will be the leaders.
Werner wigger is particularly optimistic about quantum computing and the application of cloud computing in space. He believes that AWS has the ability to provide advanced cloud technologies that are affordable, accessible and understandable to all. This year, AWS has made quantum computing available to everyone for tens of cents. In 2021 and beyond, space will be the most advanced area of cloud technology, including food supply, tracking glacier retreat and other space application fields, which will depend on the cloud technology of space operation.
The following is a summary of the interview:
Reporter: how does AWS view the future opportunities in the field of Internet of things? What new investments will be made in the future?
Werner: AWS will continue to invest in the field of Internet of things. In the past, when we talked about the Internet of things, we would talk about it as a separate field from other computing or it technologies. But now the boundaries between these different technology fields are becoming more and more blurred.
AWS provides AWS Greengrass service for IOT. For example, in mines, in order to reduce the risk of production environment, we will have driverless trucks, and in more and more factories, the way they use sensors has changed. In the past, sensors may be used to find products with quality defects in the assembly line or send an alarm, but now more and more sensors are able to collect and provide data flow to avoid false alarms, and to find products with quality defects more accurately. These things around us have become active components of networking.
For example, when talking about smart home, we feel that we should perceive through a screen to realize the automation of the family. But in fact, we think that in fact, every device and every part of the family has been activated and become an active thing. I used to use my mobile phone to check how much electricity is consumed by my home devices, but why not be more intuitive? For example, when I see the controller on the wall changes from green to red, I know that the energy consumption of this part is relatively high.
The variety and quantity of Internet of things devices are more and more popular, prompting us to consider another factor, namely the security of the Internet of things.
Reporter: this year, a remarkable event is that ARM architecture chip is widely used in desktop and data center. Apple has launched M1 chip notebook. What do you think of the industry’s accelerating trend towards arm based chips?
Werner: actually, for a long time, most of the devices around me used ARM processors.
I feel that any kind of technology development and innovation is driven by the quantity, and the quantity must come up, and the ability of ARM chip is constantly improving. In the middle of this year, AWS released two arm architecture chips, which improved the performance price ratio by 40% compared with the previous CPU.
Intel will not win more market share in the data center, although Intel will win more market share in AWS 86? Just like a decade ago, Intel beat IBM on server chips.
Werner: in fact, I think arm’s momentum is not going down. From the data point of view, we can still see the strength of x86 architecture in computing and storage, but when looking at general-purpose computing, arm has many advantages to explore and utilize.

Reviewing the two arm architecture chips released by AWS, graviton1 and graviton2, have achieved great success in terms of customers. We can see that many customers are migrating to arm architecture. But if you want me to predict whether such a time will exceed x86, I find it difficult to accurately predict.
Why do you say that? Because the migration from x86 architecture to arm is not simply to simply take things over and move them over. It requires customer input, and many applications need to be migrated from the original x86 architecture.
If there is one company in the world that can make such a big investment in arm architecture, it is AWS. We invest in this area more from the perspective of customers to help them invest and innovate in this field, and our requirement is to continuously reduce the cost and improve the performance.
Reporter: IBM, Google, Microsoft and Alibaba are all developing their quantum computing technologies. How does AWS compete with the above manufacturers?
Werner: actually, we don’t think it’s a competitive relationship, because quantum computing is actually in a very early stage of development. The purpose of AWS providing Amazon bracket technology is to enable people to obtain the hardware and software of these quantum computing. We are all exploring such a technology. Different manufacturers have different types of hardware and software in quantum computing.
From the perspective of AWS, we mainly use the scale effect to reduce the cost of quantum computing.
When it comes to the popularization of quantum computing, we should realize equal rights in technology. When it comes to general computing forms, the first thing that comes to mind is that there are various programming tools. Usually, developers will be provided with a variety of development tools to choose from, and each developer will have its own number. But in the field of quantum computing, because it is still in the early stage of development, these can not be realized, so we need to promote the development of such a cycle.
Now for quantum computing, we need to find a suitable way for quantum computing software to truly release the power of quantum computing.
Reporter: a relatively large application field of quantum computer in science and technology industry in the future may be in machine learning and artificial intelligence. What kind of expectation do you have for the help of quantum computing in AWS business? Scientists at Google have predicted that machine learning may run on quantum computers in the future within 10 years. Do you agree with this?
Werner: speaking of that, I certainly hope I have a crystal ball in my hand that can see the future and tell you the answer. Maybe not from the perspective of quantum computing, but if we want to make a prediction of quantum computing, I think it is too early.
Is it possible to do machine learning on quantum computers better, more powerful and faster? Maybe. For example, from the point of view of data analysis, it would be very helpful to do it through quantum computing. I can give you a prediction that the next ten years will be like this and that, but looking back at the present ten years later, I don’t think it is necessarily the case.
When it comes to machine learning, it is a very comprehensive and diverse category, which includes different types or processes of machine learning. What are the key points? You need to speed up the training part of machine learning. What is the final output of machine learning? It’s a different model. That’s why AWS has specially launched the special AWS inferentia chip for deductive reasoning, which can greatly reduce the cost of work.
Our role now is to continuously introduce some special chips, such as chips for different machine learning processes or types, so as to reduce costs and improve performance.
Reporter: Jeff Bezos owns blue origin (a private space company founded by Amazon founder Jeff Bezos in 2000). Does AWS plan to further cooperate with them?
Werner: Blue origin is one of many customers for AWS. We treat it like we treat SpaceX without special treatment.
Reporter: Based on China’s post epidemic performance, there are several areas that are more affected after the epidemic, including automobile, retail, finance, export, hotels and tourism. In the future, which of these industries in China will be more easily exposed to the cloud, or in the cloud era, which of these industries will develop more rapidly based on Cloud Computing?
Werner: I can also add that more industries, such as entertainment, media, aviation and so on, will embrace the cloud more after the epidemic, and use cloud services more quickly, because there have been great changes in the supply and use of services. But from my point of view, these large enterprises take advantage of such a crisis and turn it into an opportunity. They used this period of time to do a very profound reflection, determined to use technology to achieve real industry innovation, to push their industry or enterprise to a higher level.
For example, there is a very large enterprise in Amsterdam in Europe. They are affected by the epidemic, and they are determined to put 100% of it facilities into the cloud.
A very important reason for many enterprises to go to the cloud is security. Especially during the epidemic period, for many large-scale financial service institutions, medical service institutions and even life sciences, the only way for their customers to obtain services is through the Internet or digitization. It is such a situation that these large enterprises finally realize that it is not enough to protect their own security in the traditional way.
Reporter: on the premise of data security, breaking the information island is the future trend. How does AWS plan this?
Werner: AWS has a lot of customers, so we have to protect their data security according to their different situations. In many cases, machine learning technology is used.

One is Amazon Macie, which mainly guarantees the data security from the perspective of data access, that is, according to the different types of customer data, it ensures that which data is of high security level and is protected first.
The other is Amazon guard duty, which analyzes a large amount of data, summarizes the rules, quickly finds some abnormal situations, and protects the data from this aspect.
In terms of information island, many customers have both resident database and centralized data lake on AWS side. At this time, we will start from the sound identity of customers. We have a special datalake formation, which allows customers to verify their identity and access management, and protect them according to their identities and whether they have access to data.