Amazon Cloud Technology: "Cloud + Data + AI" Helps Enterprises to Transform Digitally and Intelligently
Article from qubits
The originator of e-commerce and retail giant Amazon, it's time to add a label.
From the latest 2022 Q1 financial report, the cloud computing business has become Amazon's main source of growth and profit:
In the first quarter, operating income increased by 37% year-on-year to US$18.44 billion; operating profit increased by 57% year-on-year to US$6.52 billion; profit margin was 35.3%, which was higher than the 29.8% in the previous quarter and 30.8% in the same period last year.
You know, Amazon is a giant in the global cloud computing market, with a public cloud market share of 38.9%, more than the sum of the last three.
The large volume and high growth rate have even led to the emergence of the " growth myth " in the market.
Gartner 2020-2021 IaaS public cloud market share data
From another perspective, Amazon Cloud Technology is also ranked in the Leaders quadrant in the Magic Quadrant™ for Cloud AI Developer Services report published by Gartner®.
To discuss what is behind the "growth myth", innovation must be mentioned.
Especially in recent years, cloud computing, as an infrastructure that provides AI with large computing power and big data storage and transmission capabilities, has also begun to deeply integrate with AI.
Jumping out of the scope of cloud computing, Amazon Cloud Technology has also been rated as the first in innovation in all domestic AI development platform application markets.
Frost & Sullivan & Head Leopard Research Institute "2021 China AI Development Platform Market Report"
This aspect of Amazon cloud technology has always been easily overlooked in China, just because cloud computing does not directly contact consumers, but provides support behind every aspect of daily life.
This change in financial data will serve as a striking sign and will become the starting point for the market’s perception of Amazon to flip.
For a long time, Amazon Cloud Technology will be seen by more people in promoting the implementation of new technologies in China, helping Chinese companies go overseas and localizing foreign companies, and promoting the digital and intelligent transformation of traditional industries.
New technology landing
Smart cars are definitely one of the hottest industries to talk about at the moment, but it is difficult to land on the road of autonomous driving alone.
According to research conducted by the RAND Corporation of the United States, if an autonomous driving algorithm wants to reach the level of human drivers, it needs at least 17.7 billion kilometers of driving data to improve the algorithm.
Not to mention the boundless matter of reaching the human level, according to the most recognized SAE automatic driving classification standard, reaching the L3 level requires a road test mileage of 20 million kilometers.
The scale of data generated by tens of millions of kilometers of road tests must reach the EB level, that is, the common TB multiplied by 1024 and then multiplied by a 1024.
What is more troublesome is that the data formats required by different links are not uniform. For example, data import requires S3/NFS format, data preprocessing requires HDFS format, AI training requires NFS format, and simulation and model verification follow. . . .
From the High Performance Computing Professional Committee of the China Computer Federation, "Data-intensive supercomputing technology white paper"
Many start-up companies in the smart car industry have limited human and material resources. If they want to build their own IT systems, they can't do anything else. On the one hand, the scale of self-build is difficult to keep up with the pace of the fast-growing business, and on the other hand, it is operation and maintenance. It takes a lot of energy.
In order to focus on core technology research and development, migrating to third-party cloud computing services has become a natural choice.
Domestic autonomous driving companies that use Amazon's cloud technology machine learning platform include Momenta, Zhijia Technology, etc., and globally there are Aurora, Mobileye, and TuSimple.
Providing individual cloud computing resources to each customer is not the entire consideration of Amazon Cloud Technology.
After accumulating a lot of industry experience, Amazon Cloud Technology launched two targeted new services in December 2021: Amazon IoT FleetWise , which solves data collection problems, and Amazon for Automotive , an industry solution.
It also launched the Amazon SageMaker Canvas no-code machine learning platform for people with zero machine learning experience.
Amazon SageMaker Canvas visualizes many steps of a machine learning model as an interactive UI, allowing business, human, and financial departments to quickly generate machine learning prediction models and solve problems at work without writing a single line of code.
BMW has already adopted AI across the entire value chain, enabling it to create added value for customers, products, employees and processes. Amazon SageMaker Canvas powers the expansion of AI/ML across the BMW Group. With SageMaker Canvas, its business users can easily explore and build ML models to make accurate predictions without writing a single line of code. Amazon SageMaker also enables BMW's central data science team to collaborate and evaluate models created by commercial users before releasing them into production.
Chinese companies go overseas and foreign companies localize
AI and data analysis are now not only patents of cutting-edge industries, but also useful in the wider consumer and Internet industries.
Among them, intelligent manufacturing and the digital economy going overseas have broad prospects under the double dividend of capital and policy.
Compared with the autonomous driving industry, there are several more troubles faced by overseas enterprises. They need a unified global infrastructure structure, face the risks brought by cross-border payments, and meet the increasingly stringent data security compliance requirements of various countries.
As a result, because of its own positioning, Amazon Cloud Technology is especially favored by Chinese companies that want to go overseas and foreign companies that want to land in China.
In iResearch's statistics on China's public cloud market including overseas business, Amazon Cloud Technology ranks second.
An example of Chinese companies going overseas is OPPO , whose overseas shipments of smartphones account for more than half, and the wearable smart device market is also under development.
OPPO's AI Xiaobu assistant has a monthly life of over 100 million. How to reduce the cost of AI inference and improve the efficiency of AI inference is the key.
In addition to finding ways to optimize the algorithm by yourself, you can also hire a foreign aid, a dedicated inference chip.
OPPO finally chose to deploy Xiaobu Assistant on the Amazon EC2 Inf1 instance, using Amazon Cloud Technology's self-developed Inferentia inference chip , which can reduce the single inference cost by up to 70% compared to the previous generation GPU-based instance.
In the two scenarios of Q&A and chat, Xiaobu Assistant can save up to 35% in overall reasoning, and reduce end-to-end latency by up to 25% .
Migrating to the new chip is not a lot of work, and requires minimal code changes when paired with the Amazon Neuron SDK.
The representative of the localization of foreign companies is Daniel Wellington , a watch and jewelry manufacturer from Europe.
After entering the global market, they found that a common problem was the time difference between the headquarters and global consumers.
For example, when a customer applies for repairs or returns and the person in charge of the review is still late at night, they have to wait until the next day. The consumption experience is very bad.
They later created an automated process based on the Amazon Rekognition Image Recognition API, and image recognition-based returns were 15 times faster than before.
From this small case, we can also see that cloud computing has a greater room for development.
Compared with the few enterprises that have their own AI technology and need computing resources, more enterprises need to customize their own business processes based on their existing AI capabilities.
In this regard, in addition to the Amazon Rekognition mentioned above, Amazon Cloud Technology also provides a series of related products.
Amazon Personalize , which presets the necessary infrastructure and algorithms for the recommendation system, provides an API interface, and can quickly build a personalized recommendation application. Lotte Mart uses it to increase the number of products that customers have never bought by 40% .
Amazon Connect brings AI agent scheduling, risk fraud detection, sentiment analysis and other capabilities to the contact center, which can save up to 80% of the cost compared to traditional contact center solutions.
Amazon Lex , which opens up the same technology as the Alexa voice assistant, can build, deploy and manage customized voice chatbots, and can natively integrate the contact center built by Amazon Connect.
Using these products does not require professional AI knowledge, and only traditional software developers can quickly build AI applications.
But what about companies that lack even traditional IT development capabilities?
Intelligent transformation of traditional industries
Digital transformation and intelligent upgrade are hot words in recent years.
According to the forecast of IDC, an authoritative market research organization, AI will become an indispensable part of all enterprises by 2024.
With the emergence of more and more traditional industries such as industrial manufacturing, logistics, energy, transportation, agriculture, etc., the demand for intelligent transformation of more and more traditional industries will burst, and 25% of AI investment will be in the form of Outcomes-as-a-Service. to drive innovation.
For manufacturing, an important use case for AI is demand forecasting.
In particular, the capricious epidemic has caused unprecedented fluctuations in customer demand and upstream and downstream supply chains in the manufacturing industry.
Foxconn partnered with Amazon Machine Learning Solutions Lab to develop an end-to-end demand forecasting model for one of its factories using the Amazon Forecast time-series forecasting service.
The solution improved forecast accuracy by 8 percent , saving the plant $553,000 annually .
Amazon Monitron also provides end-to-end AI capabilities. The basic usage is to monitor anomalies in industrial equipment, and more advanced is to use machine learning to detect real problems before the equipment occurs.
The professional term is called " predictive maintenance ". On the one hand, it can prevent the operation of the entire production line from being affected by the unexpected shutdown of a certain equipment, as well as possible safety problems. On the other hand, timely maintenance before failure can also increase the service life of a single equipment.
This solution has been used in various industries such as musical instrument manufacturer Fender and GE Gas Power , a natural gas power generation supplier of General Electric.
The integration of cloud, data and intelligence is the answer of this era
Having seen so many cases from all walks of life, it is not difficult to conclude two laws.
First , from high-tech to the Internet, consumption to more traditional industries, it is inseparable from the close integration of cloud computing, AI and data analysis.
Data is often compared to "digital oil" and is the most important factor of production in this era. AI algorithms are needed to mine more value from massive data, and AI algorithms need cloud computing to provide a lot of computing power.
This is the main logic behind the growth of the global cloud computing market from 0 to $705 billion in 2021 in 15 years.
Second , industries that are farther away from technology have greater demand for intelligent upgrades and more complete, end-to-end solutions.
In this way, the integrated development of big data, artificial intelligence, and cloud computing is the answer in this era .
So during this period of time, we have seen that the original big data centers in various places have been upgraded and transformed into intelligent computing centers. Companies that started with AI algorithms, such as SenseTime, have begun to build their own computing power supply systems, while cloud computing companies are developing AI chips and Develop AI technology.
Among these many players, the use of comprehensive solutions, comprehensive AI/ML tools, MLops methodology and services to lower the threshold of AI use is the advantage of Amazon Cloud Technology and the driving force behind the "growth myth".
Judging from the performance of Amazon Cloud Technology in the domestic market, its technical system has not had the problem of acclimatization.
So many cloud services mentioned above are actually just the tip of the iceberg in this technology system.
Every year, Amazon Cloud Technology will launch thousands of new services, including 250+ AI-related services.
Such a number makes relevant practitioners feel like "you slow down, I can't learn it".
Fortunately, Amazon Cloud Technology will organize and present the recent trends in the form of the INNOVATE conference every year.
It is reported that this year's INNOVATE conference is based on the theme of " New Engine of Artificial Intelligence ", free registration and online participation.
To learn more about Amazon Cloud Technology's AI side, check it out next week.
Click the link to register
 "2021 China AI Development Platform Market Report"
**粗体** _斜体_ [链接](http://example.com) `代码` - 列表 > 引用。你还可以使用