Huawei Rotating CEO Eric Xu Speech on Responding to the Fast-Changing World
Oct 20, 2017
At the 2017 China Management World Forum & Kingdee User Conference held on October 15th, Huawei Rotating CEO Eric Xu delivered a speech titled "Responding to the Fast-Changing World".
CEO Eric Xu delivering his speech at the event
Distinguished Guests, Ladies, and Gentlemen,
I am honored to attend this event. Thanks to Chairman Xu Shaochun for inviting me to this event. We are living in a fast-changing world. We need look no further than in front of us to see the changes taking place in society. I will be sharing on what I think we need to consider when responding to the fast-changing world.
Living in a Fast-Changing World
Consumer, enterprise, and public agency requirements are changing in China and the changes in tech are rather dizzying. In just the recent five or six years, many new words have pervaded our lexicon, such as cloud computing, Big Data, AR, VR, automated driving, artificial intelligence, and so on. At Huawei, we are most concerned with the overall tech strategy. Yet, getting all our managers to understand what these words mean is no easy task let alone applying these rapidly updating technologies to our daily operations and adapting the tech into the products our customers demand. Business models are also changing. Changing business models are often the key for us to compete in other races. For example, in one race, we may need stamina and in another we may need speed. It is precisely because Kingdee moved to undertake cloud adoptions one step ahead of others that it is experiencing the growth that it is today.
Each enterprise must decide how to respond to this fast-changing word and solve the challenges before them. Looking at the development of Chinese enterprise over all these years, many of them competed internationally by copying others. Today, however, the "me too but cheaper" approach in no longer a viable path, it simply won't work because changing requirements must be dealt with immediately. Innovative technology and new business models must be adopted to create new roads. Taking this approach allows us to compete with established competitors on a different track. Our strides might not be as big as others at the start of the race, but we will build speed as we develop in the future.
Speaking to certain industries like manufacturing, we need to pay attention to the upgrade in consumption. Consumers today are not like those of the past who were mainly concerned with price – if it was cheap and could do the job then done deal. Consumers today want name brands, quality, and customization. If you want to build a brand, you need innovation and you need to provide customers with better quality products and services. This requires investment into R&D. I have talked with many entrepreneurs, founders, CEOs. In building R&D capabilities, the biggest challenge is in the conflicts with conventional thinking and the embedded business logic because investments into this arm cannot produce immediate gains. It's impossible to get returns from R&D investment within three to five years. Especially when it comes to research, it would be a bit unreasonable to expect profitability within a five-year period. Looking back on the roads Huawei has taken, we made continuous investments for each product we added and for each industry we provide products for, and we did not start making money for quite some time in most cases. We ran some stats and found that it took us eight years on average to reach the break-even point. This eight-year timetable on realizing gains would be hard for most enterprises to digest in their investment strategy. However, we have come to understand completely that only though continuous investment can we stay ahead in the industry and become a leader and eventually become the world leader.
As a provider of software and applications, we cannot be bound to the development, product supply, and business models of the past. The present approach in developing products and providing services over public cloud (including AI platforms) and moving all software and applications to the cloud in 'as-a-service' convenience for customers is the ideal method that has been proven to work. Kingdee has been resolute in making the transformation to the cloud quick, which has paid off in rapid growth over the last couple of years. It is also because of this transformation to the cloud that enterprise like Huawei can offer Kingdee Cloud services to our device retailing with the agility to grow as customers expand. Otherwise, adding one node at a time would make it difficult to respond quickly to demand.
The approaches in integration are not the same as those integrators adopted in the past. Development hinges on the ability of the integrator to migrate and consolidate into the cloud. If only conventional approaches in integration are taken, it will be hard to secure victory in the future.
Artificial Intelligence – the New General Purpose Technology
Next, I will talk about AI, which has become the new general purpose technology (GPT). This is becoming more of an accepted fact. In his book titled Economic Transformations: General Purpose Technologies and Long Term Economic Growth, Richard G. Lipsey argues that socioeconomic development depends on the continued promotion of GPTs. He defines GPT as a product, process, or organizational type of technology. According to the latest edition of Wikipedia, economists cite that humanity has adopted 26 GPTs in the course of its development and AI is on that list. GPTs are essentially something that can be used for many purposes and applied to almost every area of the economy at huge benefit. AI helps solve those issues with existing solutions in a more efficient manner while also solving many problems for which no solution previously existed. Having mindsets that enable us to adopt AI tech to solve existing and future problems is what will determine if we are able to take a leadership position in the competition and will be the key to establishing advantages.
AI does not have a set definition that everyone accepts as accurate yet. Different organizations have different emphases and positions. Gartner considers AI to be a rather inclusive and comprised of many other technologies. Narrow/soft AI does not contain emotional awareness and is focused on narrow task sets. Examples include Siri and Kingdee Little K. Strong/real AI relates to robots able to exhibit the general knowledge workers need to complete tasks and show expressions and perform actions resembling those of humans. This is a distant goal. What most in the many industries are focused on right now is machine learning.
AI experienced two winters from the 70s to the early 90s. Academia even labeled those researching and disseminating AI as deceivers. For a time, the word AI was replaced with Big Data. Now, AI is being used again all of a sudden and it seems mention of Big Data is waning. China has turned Big Data into its own industry and the information industry has become known as the Big Data sector, but this is another case in and of itself. The changes in AI started back around 2012 when Alex Krizhevsky, Doctorate candidate at the University of Toronto and his colleagues were able to use a deep neural network to reduce the error rate on ImageNet down to 15% from the previous 30%. This achievement leads to a revival in AI and it seemed nearly everyone in industry to academia were once again taking about it.
Yet, almost everyone from the layman to the expert look at AI in different ways. Some are scared of it. Hawking is most scared of AI, he even feels black holes are nothing fear but is terror-stricken when it comes to AI. Many others are optimistic and want to accelerate its advancements. So, some are afraid and others are full of enthusiasm. In 2012, U.S. President Obama released a white paper titled Artificial Intelligence, Automation, and the Economy in which he argued that, while robots are unlikely to match or exceed humans in general ability or intelligence in the next 20 years, they would meet performance or exceed the performance of humans in more and more tasks. His general perception is that it would be hard for robots to fully take on human attributes and be just like humans but that robots will be able to match or exceed human performance in some narrow task sets. He feels this is a safe assumption for the next two decades of development in AI.
In all reality, the mechanisms involved in AI are simple, which is also why there are many types of applications for each element. AI depends on large volumes of data, algorithms, and 'break-in learning' to develop a model. Once the general model is established, keying in command A will produce response B, with different entries producing different responses. Such processes can involve anything from labeling an image, checking credit for loan approval, implementation of precision marketing with online ads, voice recognition, and so on. Although such machine learning modes are hard to explain, the outcomes are what they are and there has been much success with widespread application in many domains. And, the area of those adoptions will continue to spread. So, we must be objective and see how AI is changing all industries and even every enterprise. Enterprises should not just look at the various types of AI technologies, they need to rethink how the new tech will impact and change the makeup of the functional departments in the enterprise in everything from sales to marketing, from financial management to resource management, and everything to do with business. I was happy to see from CEO Sun's presentation on the Kingdee Cloud to involve moving all functional departments to the cloud. I also hope Kingdee will take some early steps in AI and apply it to each domain at the enterprise to secure another top spot in the industry.
We must also rethink each industry and how AI will impact each sector, even to the point of totally disrupting certain ones like automotive. Self-driving electric vehicles are expected to account for 16 trillion in value in automotive, including peripherals. This will be a huge shake-up. AI also has the potential to disrupt the medical and education industries. Professionals in each industry must consider how AI will impact them. They must ask themselves if AI will disrupt their industry and which new models they need to reshape their industry, which is also something we need to think about in the future as well. We must also reconsider the interactions between humans and machines, how they will interwork in the future, and how man-machine interfaces need to be designed. Although AI is still in its early stages of development, it has become a GPT and it will bring huge changes to enterprises and each industry. The extent of these impacts should urge us to rethink our views.
Understanding the concept in terms of cell phones is likely the easiest way to grab hold of the dynamic. In reality, smart phones are not that 'smart' but they are becoming super phones able to understand us, become aware of our preferences, actively respond to our requests, and even remind us of schedules and events. The super phone is now a 'must-have'. We did a little experiment in the last quarter of last year in our launch of the Magic super phone that moved beyond standard smart phones.
AI will also change each organization. The organizational framework in the past for enterprises entailed a triangle-like scheme with the leaders and experts at the top, key workers and entry-level managers in the middle, and average workers at the bottom. The organization in the future will be more diamond shaped with much of the repetitive work handled by AI applications and robots. Finance, for example: if Kingdee Little K was able to complete many of the finance tasks, accounting clerks would not have much to do. Gartner predicts that in 2018, over three million people in the world will be directed by robots. Didi and Uber drivers are being directed by machines as we speak.
AI will also change each career. Looking at the impact analysis on jobs in the USA, McKinsey points out that 30% of the tasks in 60% of the jobs will become automated. The figure is different for China but we can be sure that many of the important jobs will become automated and even some of the more professional jobs requiring intelligence and even a small portion of the jobs needing human thought will be replaced by machines or involve human-machine interaction. That means some careers may be eliminated but we should not be afraid of that because AI will also create many new careers. Data lies at the heart of AI and there will be huge demand for data analysts at enterprises and in each industry and data model engineers will be needed to 'teach' the machines. So, some jobs may disappear but new jobs will be created, which is how society has always advanced.
Of course, AI does have many limitations. AI is only adept in performing narrow tasks. It is better suited to simple tasks and not engineering-based developments or implementations requiring extensive experiments and personnel with abundant experience. The new pillar tech also requires lots of preparatory data that must be accurate – the more accurate the data, the better the AI will perform. If the data is inexact, the AI will go off the erroneous data in performing its task and produce a non-desirable or even harmful outcome. Since the working principle of AI is still inexplicable, it's challenging to apply AI to particular key tasks that require predictable results.
The new round of development in AI has just begun. There is still plenty of room for improvement in terms of the technical capabilities. Allow me to share some more of my thoughts on how enterprises can identify and effectively implement AI. First, in terms of technology, the algorithms for AI have existed for a while now. Even the framework algorithms for the deep neural network have not changed much since the 1980s. So, why have there been so many changes in recent years? The short answer is enhancements in computing capabilities and the huge volumes of data collected. These advancements have promoted AI. Developments in AI will continue to rely on compute power in the future as well and will place requirements on the expandability of the algorithms, so the algorithms will be critical as will computing capability. Another point is that leaders of enterprises usually want to use AI for existing tasks to improve productivity and realize better levels of automation to alleviate human workloads. We need to open up to using AI to solve some of things we are not sure of or solve some problems without existing solutions. So, leaders and entire staff at enterprises need to bring AI concepts into how they work and think about how to use the tech in improving existing processes and solving unknown problems. The trajectory of this tech makes personnel in the AI field a hot commodity. Finding these professionals is one way to satisfy the need. We feel a more effective approach is to train our own personnel. Equipping them with the needed capabilities and concepts so they can then think about how to apply AI to solve problems, integrate the tech into their tasks, and continuously learn and improve as they resolve issues.
At Huawei Connect 2017, we launched an enterprise intelligence platform featuring a one-stop AI service platform with a rich store of fine-grained APIs able to work with the various algorithms for specific industries and with heterogeneous computing infrastructures, allowing our customers to use a rich store of AI utility to solve their problems. We are working hard to improve on the platform, fueled by our mission to enable all enterprises to use AI in their functional departments and in the product offerings with greater ease, agility, and convenience.
Yu Chengdong has already released our mobile AI strategy when it comes to cellphones. We are aiming to make Huawei phones a personal digital assistant for users. Huawei cellphones will become more user aware and tuned, moving far past the talk and internet browsing functions common today. Our phones will be geared completely to users in the future. Yu Chengdong also unveiled the world's first smart phone chip for dedicated neural network processor: the Kirin 970 featuring strong advancements in AI functionality over the series of chips developed for Mate 10 with many times the image recognition capability than before.
Huawei Cloud – Driver of Enterprises Growth Today and Tomorrow
We started to invest strategically into the IT industry back in 2008. Before that, we focused almost all of our efforts in telecommunications. Our biggest investment back then was cloud computing because that technology offered us the opportunity we needed to break into the market with the advancements in the tech. That was our prediction in 2008 and it was a correct assessment. We have produced many achievements. In 2009, CEO Ren delivered an insightful speech on how cloud computing would disrupt the entire IT industry, including the telecom network. Cloud computing has developed quickly over the last several years. According to a report released by IDC in 2017, Huawei was number one in the China market in government cloud and the Big Data market. We have performed well in providing unified Big Data platforms and cloud operating systems to our customers as evidenced by our strong market share.
Huawei Cloud is our company's brand when it comes to cloud. Huawei uses cloud computing technology to help large enterprises ramp up their cloud adoptions in IT infrastructure and to deliver the Big Data analytics they need. We have added an operations system on top of the vetted technology in the public cloud to allow enterprises to conveniently purchase cloud services. We also added an O&M system so users can get a quick response for any problem they may encounter. We are upping our investments and developing public and private clouds in tandem to provide users with hybrid cloud solutions able to meet their requirements in operations and maintenance.
In terms of the cloud, our offering adds Internet of Thing (IoT), Big Data, and AI services to the basic framework comprising compute, storage, networking, and security to provide customers with a full stack of solutions. The Huawei Cloud is not just software and services, it includes chips, hardware, and a complete ecology able to combine on- and off-premise resources to solve customer problems and help enterprises execute their IT transformations. Huawei Cloud is also able to support Kingdee in better serving all its customers in provisioning of software and applications. Huawei public and private cloud are developed in tandem so they use the same frameworks and APIs, which means applications can be easily migrated across both profiles. Applications developed over the Huawei cloud can also be moved across the public and private clouds at all types of organizations and then back again. Software and application developers can deploy their offerings in both private and public clouds.
Moving now to the IoT. Perhaps you have heard this word thrown around quite a bit but have not thoroughly understood it or its value. IoT offers plenty of solutions to issues that previously could not be solved. In the IoT ecology, we have more than 50 success cases with our solutions. The new tech makes many things possible. For example, environment testing, monitoring of production safety, and many other scenarios benefit greatly. Huawei is an enabler of IoT, helping partners solve problems in each industry and create value with the capabilities of our IoT connection platforms and chips. IoT and the cloud are here, and combining the two will help us solve many problems we need to solve yet have no solution for at present. The new capabilities will create value for customers and enterprises.
We consider the public cloud to be more than just Internet applications. It contains the tech for long-term enterprise agendas and can store a full house of experience and content. Public cloud is able to power R&D and new innovative capabilities with its makeup of ICT infrastructure, AI, IoT, and whatever combination of technology and products are required for the operation. Everything is provided to customers in 'as-a-service' formats. Take the business model of online-only phone makers as an example. Their products are no different from those of other makers, and the only difference is that they use the Internet to rapidly sell products to users. I don't identify with the 'Internet+' idea. Huawei has continued to emphasize the '+Internet' concept to gear up for the future in China considering the largeness of industry in the country and the huge volumes of data. We created the Honor brand in our cellphone lineup that adopts the '+Internet' sales model. Huawei Cloud also uses the Internet to provide our end users with the solutions culminating from our many years of investment and achievements in cloud-based platforms. Not every provider of infrastructure will be able to survive in the public cloud market.
China is developing quickly. More and more problems require solutions from technology. As an enterprise in rapidly changing times, it is extremely important to find ways to compete in China and on the international stage and in races dominated by established players by providing customers with solutions with different models. For us, that means using new technologies and concepts to solve long-standing problems and improving efficiency in solving these problems with higher levels of automation to help alleviate large labor inputs. At the same time, we advance using new tech and ways of thinking to solve unsolved and unknown issues. This will make us a leader.
Compared with western countries, China lags behind in some technology domains, leads in others, and goes head-to-head with them in the rest of domains. China is no longer just a laggard, and it will only move faster in the future.
We are investing and combing technologies and platforms into connection + cloud solutions. Partners are using new business models powered on Huawei's leading technologies and platforms to gear up for the future. So, what don't we do? We don't touch data and we don't touch applications. Today, we have added another "don't", which is we don't make equity investments. In the cloud domain, we don't invest in integrators or application developers. We don't have a favorite we cultivate to go compete with our partners. We provide all our Huawei Cloud partners with an open, fair, and equal environment. We are aiming to be the driver and long-term partner for Chinese enterprises in their move to the future. Our goal is to band with our partners and provide a stable, safe, trustworthy, and evolution-ready cloud service portfolio as we find solutions for the future and satisfy fast-changing customer requirements together. We are using new technologies and business models to enhance our competitiveness and establish our leading position to achieve ultimate success.
Thank you!