Shulou(Shulou.com)06/02 Report--

Human imagination of technology has always been a very interesting thing. We can see a world full of gears and steam in Victorian fantasy novels, but people are still lighting with kerosene lamps. In some science fiction works in the 1970s, there has been a prototype of portable PC, but these computers still use black-and-white screens.

The reason for this interesting dislocation between reality and expectation is that we can only see the progress of one or two technologies, but we did not expect that the development of science and technology must be a comprehensive promotion of a series of technologies and even cultures.

The same is true of the AI we are talking about today. Most of the time, our understanding of the development of AI technology is biased. When we look at the recent emergence of inspired new algorithms, or speech recognition accuracy has improved by more than 0 percentage points, we often forget that algorithms are not the only protagonists in the AI technology arena.

Like the development of computing power, it is an important force to make the AI blueprint from expectation to reality.

From the development of Wisdom Computing in the past year, we may be able to try to explore the running pace of AI computing.

The Secret Code of Chinese Computing Industry: "1BI 60je 70"

The wind starts at the end of Qingping.

In 2018, a group of "mysterious codes" consisting of three numbers has emerged, which may help us understand the current computing industry in China. This set of data is "1BI 60J 70"-No. 1, 60EB and 70%.

The number "1" represents the vitality of China's computing market.

In the past year, Tide has made a series of firsts in the Chinese market: server market first, AI computing market first, cloud computing government cloud market first, Internet market first, storage market growth first.

At the same time, last year's overseas business also increased by 3.5 times, entering the top three of the global server market.

This series of first, not only the performance of the wave, but also China and the world for AI computing, cloud computing and other services show a strong desire.

"60" is a common opportunity facing the entire computing industry-the amount of data that is breaking through the tipping point.

According to the data provided by Chao, the data capacity of China's large-scale data centers has reached 60EB in 2018, and this number will reach 200EB by 2020. With the closer and closer integration of AI and the real world, more and more sensors are being embedded around us. Intelligent car terminals that record position and speed every minute, production lines that automatically view every piece of raw material, and intelligent security cameras that transmit data to the cloud at any time.

Before this amount of data overwhelms us, the computing and storage industry is facing not only a vast market, but also a new proposition in the development of technology.

This also leads the code to the last number, 70, the important impact of computing technology on the development of AI.

The computing power of AI is not only responsible for the transfer of data up and down. AIStation, the management suite launched by the wave, can quickly increase the utilization of users'AI environment from less than 20% to more than 70% without adding hardware.

This means that when the AI computing task is becoming more and more complex and huge in the future, the computing industry represented by the wave has the ability and obligation to help the hardware system play a more efficient role through technological innovation, rather than blindly increasing the cost burden of AI technology by increasing hardware.

From the mysterious code of the tide, we can interpret three key messages:

1. With the development of AI industrialization, the computing market is entering an unprecedented explosive period.

2. With the rapid increase of data processing requirements and data volume brought about by AI, new requirements for computing technology are put forward.

3. In addition to the hardware product itself, computing technology has more far-reaching impact on the landing of AI.

From Linear to Tensor: the third-order Leap of Computing Industry

These three messages all point to the same problem. In the future blueprint of AI, the role of the computing industry is very different from that in the past. The previous forms of research and development, assembly and sales of servers can no longer meet the requirements of AI for computing power, and the computing industry needs to achieve self-definition through more technological innovation.

The difference among them is the leap from linear to tensor, which needs to break through not only numbers, but also different spatial dimensions.

From the tide, it reflects the attitude of the expansion of the computing industry dimension. We can go up step by step and see how the tide is moving towards the leapfrog of the computing industry.

The first-order talent ecology

An interesting phenomenon is that when AI enterprises are going deep into colleges and universities to carry out talent programs, the voice of computing enterprises seems to be lower than algorithms and data. But from the perspective of AI applications, the requirements for developing new algorithm models in the future will be far lower than those for dealing with AI computing. In fact, we can already see the clue. In terms of relatively mature technologies such as face recognition and speech recognition, there is little difference in the accuracy of algorithm models among different manufacturers. The current match point lies in whether the applicable scenario of the technology can be enhanced through edge computing.

Whether there is a deep enough talent training plan is not only very important for the development of a computing enterprise, but also determines whether there are enough balanced talent types when AI lands in the future.

In this respect, the tide is already one step ahead of others. In the Plan T launched in 2018, Chaochao has been helping colleges and universities to cultivate a talent ecology that can better meet the needs of the future computing industry through training camps, providing AIStation application resources to colleges and universities, and participating in the compilation of courses and teaching materials.

Second-order industrial relationship

Last year, when Harvard Business Review selected the Ram Charan Management practice Award in China, Chaochao's JDM model won the award. This award represents a change in the relationship between computing enterprises and the application industry in the AI era. The emergence of AI technology has brought about great changes in the industry's demand for computing. In the past, the requirements of enterprises for data computing and storage are often not much different, but the emergence of different algorithms makes the needs of enterprises begin to be personalized. For example, some enterprises may rely more on large amounts of data, so they need to have higher requirements in terms of storage and server capacity, while others pay more attention to timely feedback of data, so they need more edge computing.

The reason why the tide of JDM model is well received is that it has changed the situation of zero communication in the computing industry in the past and transformed it into in-depth customized production according to customer needs. A large part of the reason why Chaochao achieved great success in overseas markets last year is that the JDM model can more efficiently meet the needs of enterprise intelligent transformation.

The third stage One More Thing

After completing the two steps of talent ecology and industrial relationship, if you want to take another step forward in the computing industry, you may have to provide a "One More Thing" that is detached from the existing industry matrix, just like every mobile phone release. For Chao, their One More Thing is HPC-- high-performance computing.

Going back to the beginning, when a technology is just in its infancy, it is difficult to imagine how it will change our lives in the future, because the future is shaped by the combined action of countless technologies. Perhaps the future development of two-dimensional materials will make our nerves wander with dust-sized sensors, or maybe under the action of 5G, we no longer need terminal storage, but let huge amounts of data wander in the cloud forever.

When there are countless possibilities around you, preparing a "One More Thing" is obviously an infallible choice. For the tide, the powerful computing power of HPC provides more rich technology exports and choices for future AI applications. In particular, the threshold of HPC technology is higher, and HPC, as an One More Thing, is likely to become an important moat in the unknowable future.

The past year has been a significant one for both the wave and the computing industry. Achieved great-leap-forward development, but also let the computing industry into a new era, the achievements of the past year is not only a monument, but also a starting point. What is more to look forward to is the supersonic travel brought about by the whistle of the new era.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.