China's economic development has entered a brand-new stage, and the international and domestic situation has undergone major changes.Under this background, we need to have a deep understanding of the basic characteristics of the new stage, make clear the main tasks of the new stage, and determine the main countermeasures of the new stage.《中国经济发展新阶段》 is systematic analysis and exploration of major theoretical and practical issues in the new stage of economic development, trying to theoretically sort out the internal mechanism of the evolution of the new stage of economic development, and on this basis, put forward suggestions for economic policy choice, enterprise strategy choice and scientific research in colleges and universities.

1.The choice of entrepreneurial strategy is to think before acting

As for entrepreneurship, many people advocate learning entrepreneurship in the process of entrepreneurship. Some people also think that the choice of entrepreneurial strategy is very important, any entrepreneurial idea has a variety of ways to realize, because different ways to realize not only the cost is different, the difficulty is also different, the possibility of success or failure is also different. What's more, after trying one implementation, if you find that it doesn't work, you may have lost the opportunity to try something else（Gans et al.,2018）.

Through extensive research on the law of entrepreneurship, we prefer to think before act. In fact, Gans and his colleagues present a systematic approach that involves a careful selection of four elements, namely "technology, market, competition, and identity", as well as four strategies, namely "intellectual property strategy, value chain cooperation strategy, disruption strategy, and architecture strategy".

2 .Focus on technology maturity

A large number of studies have shown that although the evolution of new technologies has certain rules to follow, such as following the shape of the S-curve, there are still great uncertainties. For example, the specific shape of the S-curve, such as a large time span, a long stretch, a rapid evolution, and a steep curve, may be very different between different technologies.

In the realistic competition of enterprises, how does technology development and product development proceed? The research of Wheelright and Clark(1992) shows that some important principles need to be followed. For example, we should separate technological invention from technological application. This is because technological inventions not only take time, effort and resources, the results are also largely unpredictable; If invention and application of technology are mixed together, it is difficult to avoid delays in product development. Taking HP as an example, it determines the key technology fields and key technologies to be mastered by the enterprise according to its competitive strategy, and develops and stores these technologies in advance. In the process of product development, the relevant technologies are actually selected from the already stocked technologies for "assembly".

A common mistake many businesses, including start-ups, make is to rush to develop products while the technology is still very immature, believing that this is the way to seize market opportunities earlier. This works when the technology is very simple, because it's easy to "hack". However, if the technology is complex and the "technology breakthrough" is not successful, the startup will fail.

Another principle is to pay equal attention to product technology and process technology, which is an important condition for producing high quality and low cost products. However, the reality is that many enterprises either ignore process technology development, or ignore product technology development. In the 1980s, many American scholars reflected on the decline in the competitiveness of American enterprises, and found that an important reason was that American enterprises did not pay much attention to the development of process technology, believing that R&D was mainly "product development" and that the production process could be obtained relatively easily. As we have introduced in Chapter 4, in the process of developing video cassette recorders, AMPEX, an American company, focused more on product development and hoped to obtain technological technology from Japanese companies through cooperation. However, it was defeated by SONY, Panasonic and JVC, which valued both product technology and technological technology (Rosenbloom et al.. , l987).

This tells us that high-tech entrepreneurship, if there are conditions, it is better to wait until the technology is relatively mature to start a business. This is especially true for individual entrepreneurs such as university teachers and students, because if they are "exposed" too early and are seen by powerful large companies or other entrepreneurs, the competition will be very fierce and the possibility of success will be greatly reduced (Rosenberg, 2010).

3. Select users carefully

Some users are willing to try new technologies and products and are not so picky about the maturity of technologies and products (von Hippel, 1988). Many users are different, only when the technology, the product is very mature, very reliable will be willing to buy. Some users are not willing to try it until many people have bought it, and some users are not willing to try it anyway (Moore,2002). For high-tech start-ups, if the technology is not sufficiently sophisticated, it is best to select users who are particularly willing to try something new, although this group may be small.

Select companies that produce complex products as users very carefully. For example, an aircraft manufacturer or a car manufacturer whose products consist of so many parts, assemblies, and systems that replacing any one of them could have a significant impact on the product, making it easy to avoid doing so, let alone using parts from an uncertain start-up. Even if this component has obvious advantages, it needs to go through a long time of trial and verification.

4. Select employees carefully

Start-ups often want to attract talent with a generous package. Our research has found that this is not necessarily the right choice. For example, when recruiting key employees, a startup insists that employees are paid less than their former employer for a period of time after they are hired, rather than more than their former employer. There is no doubt that recruitment has become much more difficult. Why do we do this? The company's answer: Key employees must agree with the founder's core philosophy, must believe that the start-up is truly promising, and must believe in its creativity, technical potential and management ability.

5 .Choose shareholders carefully

The demands of shareholders may also be diverse. Some have long-term goals, some short-term goals, some are to gain profits from the development of start-ups, and some are to gain relevant knowledge of new technologies and new markets. In the late 1980s and early 1990s, for example, there were plenty of companies with similar strength to Huawei in the telecom equipment industry, including two more in Shenzhen. However, the shareholders of this enterprise, instead of expanding its operation and improving its capabilities in all aspects, especially its technological capabilities, choose to distribute the profits after the enterprise makes profits, and as a result, they soon fall behind.

There is another story. Around 2005, a startup in Beijing is very successful entrepreneur, focus on the development of communication command system, the size of a set-top box system equipment can return a house, then introduce some new shareholders in order to expand production, the new shareholders care most about return on investment, the entrepreneur even when give college students a lecture answering the call from a new shareholders.

6 .Select technical solutions carefully

This is a very important and very complex issue. For example, digital transformation is a hot topic at present. Many enterprises take a very positive attitude towards digital transformation, but the effect of digital transformation in reality is not ideal. According to our long-term follow-up study, in the field of industrial Internet, the lack of ability of digital transformation plan to promote enterprise development may be the main reason for the unsatisfactory effect of digital transformation.

But Gridsum is different. It has unique views on some basic problems of digital transformation and a deep grasp of the basic laws of digital transformation. Specifically, Gridsum argues that for the digital transformation to be truly intelligent, it must be built on the processing of natural language by computers that can actually read books, handle both structured and unstructured data, and really think. For this reason, Gridsum attaches great importance to cooperation with universities and scientific research institutions at the forefront of natural language processing. For example, the "Gridsum Joint Laboratory for Big Data Science" of Renmin University of China was established in 2014, and the "HIT -Gridsum Joint Laboratory for Big Data Science" was established in 2015.

So could a computer in a short period of time have advanced cognitive, not just perceptual, abilities to fully understand natural language? In a general sense, the answer is no, probably 20 to 30 years or more, according to Gridsum. But there are specific areas in which major technological breakthroughs can be achieved. Based on this understanding, rather than trying to develop a one-size-fits-all, multi-scenario digital transformation solution like Microsoft, Gridsum's strategy is to "cut through the heavy stuff" and pick a specific domain or industry.

In fact, a lack of understanding of the difficulty of complex problems is enough to make a firm's competitive advantage a big challenge. Historically, IBM's attempt to create compatibility between multiple computer architectures was a painful lesson (Bauer et al., 2004). Relevant literatures have studied a series of challenges faced by digital transformation, including the complexity and diversity of digital transformation process, as well as the adaptability and flexibility of digital transformation strategy. To some extent, they have also answered why the path of digital transformation chosen by Gridsum is relatively smooth (Gao Xudong, 2021).

7. Choose your competition carefully

For start-ups, competition is crucial. Professor Bhide's research (2000) shows that it is easier to start a business in a new industry because the technology is new, the market is new, everyone has the same ability, and there is no obvious strong player. In a very mature industry with a stable competitive landscape, the difficulty of starting a business will be greatly increased, unless the industry is undergoing a great transition (Fine, 1998).

Start-ups think they are technologically advanced, nimble and youthful, and that the industry leaders are "nothing". In the case of "disruptive" technologies, there may be some truth to such judgments. But in many cases, such judgments can lead to the most fatal mistakes. Studies show that no matter how difficult innovation is, as long as the "value network" remains unchanged or the industrialization of new technologies requires many complementary assets, large enterprises may have very obvious advantages (Christensen et al.,1995; Rosenberg,2010).

8. Choose your identity carefully

Take Gridsum as an example. Mr. Liu Jianyang the CTO of Gridsum said,Gridsum is a vision-driven and strategy-driven company. I left Microsoft at the height of my career to join Gridsum because I highly agree with the core values of Qi Guosheng, the main founder of Gridsum. Gridsum's vision is to be a company like Microsoft that makes a significant contribution to society. In this vision, the company's important decisions will not be affected by too many short-term profit factors, so as to provide a suitable soil and environment for the cultivation of various capabilities, especially the cultivation of profound technical capabilities. In fact, Gridsum has had the opportunity to form a joint venture with or be acquired by a prominent multinational. In the eyes of most entrepreneurs, this is a once-in-a-lifetime opportunity. But after serious consideration, Gridsum decided that it was not in line with the company's vision and would not allow the company to grow independently, so he passed up a rare opportunity.

For example, when dealing with the relationship between expanding enterprise scale, reducing cost and enhancing enterprise capability, Gridsum seems to be more inclined to enhance enterprise capability. After successfully developing law dissecton, one option was to rapidly expand the market, scale up and reap profits. Instead, Gridsum is going to keep digging into AI technology and not give up until it's done. As mentioned above, this is a very time-consuming and labor-intensive choice. For a start-up company, it is difficult to persist in research and development full of risks for more than three years without ambitious pursuit.

In addition, in recent years, due to the general environment, Gridsum's benefits are not good, but Gridsum's scientific research has not been greatly affected, and research investment continues to increase significantly, which is also carried out under the guidance of Gridsum's "three-year strategy" (to become the leader in the digital and intelligent transformation of enterprises and government organizations). The effect is clear, and Gridsum's technical capabilities have improved dramatically. For example, in the field of big data, Gridsum is based on the advantages of distributed and parallel computing. After years of deep cultivation and meticulous work, the capability of big data products and its back-end multidimensional analysis engine technology are in a leading position in the big data industry. In the field of artificial intelligence, Gridsum has made a major breakthrough in the cognitive layer of artificial intelligence, namely natural language processing and knowledge graph. In terms of intellectual property, as of June 2020, Gridsum has accumulated more than 3,300 patent applications, including more than 2,200 invention patent applications in the field of big data and more than 400 invention patent applications in the field of artificial intelligence.