On the third anniversary of ChatGPT's launch, a strategic advice for corporate executives

If the emergence of ChatGPT three years ago was the first push that toppled the first domino, then today, three years later, we are in the midst of the cascading tremors as the dominoes continue to fall. In this wave of technological advancements that seem to update and iterate almost daily, most business managers are feeling more lost than ever: The trends are so unpredictable. How should we formulate our strategies?

On the third anniversary of ChatGPT's birth, Zhang Yu, a strategy professor at the China Europe International Business School (CEIBS), didn't follow the crowd to predict the technological trends. Instead, from the perspective of a business school professor, he returned to the essence of business and sorted out the "unchanging" underlying logic in strategy formulation in the AI era: In this era full of uncertainties, it's advisable to first identify the certainties.

There's no doubt that in the past three years, the new - generation AI technologies represented by Generative AI have brought profound changes to human society. From significantly increasing work speed (especially in entry - level graphic, text, and programming work), saving work time, reducing costs, and increasing profits, to threatening and replacing jobs (especially entry - level positions and employment opportunities for young people), we are almost daily shocked and amazed.

By now, most people would probably agree that this wave of AI technology is the "Fourth Technological Revolution" that has had a profound impact on human society, following the steam engine, electricity, and computers/the Internet.

So, why does Generative AI have such a huge impact and bring such great changes to our society, especially business activities? The reason might be that Generative AI is so far an almost "perfect" technology that can simultaneously meet the requirements of economies of scale (decreasing marginal cost) and diverse and personalized needs.

According to the analytical framework proposed by economists Carl Shapiro and Hal R. Varian in the book Information Rules ①, the prospects of technology and products depend on two dimensions: 1) Whether it can achieve economies of scale; 2) Whether user needs are homogeneous.

Previous technologies such as the steam engine, electricity, and computers/the Internet met some basically homogeneous user needs on the basis of achieving economies of scale. Although they brought huge changes to human society and business activities, there were still a large number of heterogeneous needs that remained unmet.

This wave of Generative AI technology is a relatively general AI technology. It can be developed once (with low marginal usage cost) and meet customized needs, thus completely detonating the heterogeneous need scenarios that the previous technological revolutions couldn't satisfy. That's why it has a huge impact on individuals' work efficiency and enterprises' operational efficiency and performance, and even threatens many people's jobs and employment opportunities.

Generative AI has brought us huge changes and also made people quite confused about how to formulate and clarify corporate and business strategies in the AI era. But in my opinion, the more we face such changes, the more we need to figure out which parts of the strategy remain unchanged.

As Jeff Bezos, the founder of Amazon, once said, "People often ask me, 'What will change in the next 10 years?' But almost no one ever asks me, 'What won't change in the next 10 years?' "

So, in the AI era, which "unchanging" parts of strategy formulation should we focus on? How should corporate strategies be formulated?

Focus on User Value Creation

According to the contemporary value - based strategic management analysis framework ②, enterprises cooperate with customers and suppliers to create value. The value created by an enterprise's business is equal to the customer's willingness to pay minus the supplier's opportunity cost. The magnitude of value creation determines the size of business opportunities.

According to the user value pyramid proposed by Bain & Company ③, the value that enterprises bring to customers is actually hierarchical. Taking To C products and businesses as an example ④, the first layer, which is the easiest to achieve but also the most competitive, is functional value, such as saving time and effort, simplification, and cost reduction; the second layer is emotional value, such as nostalgia, aesthetics, fun/entertainment, and symbolic value; the third layer is life - changing value, such as self - actualization, hope, and wealth inheritance; the fourth and highest layer is social impact value, such as self - transcendence.

If we use this value pyramid framework to analyze the main products and applications of Generative AI in the past three years, we can see that the vast majority of products and applications are still concentrated on functional value, that is, how to help users save time and effort, improve efficiency, and reduce costs. This also explains why in the current fierce competition among large models and computing power, most manufacturers are still focused on competing in technical parameters and cost - effectiveness, thus falling into the (vicious) cycle of "performance improvement - cost reduction - price per unit performance decline".

In contrast, the emotional, life - changing, and social impact values in the user value pyramid are more difficult to achieve, so the competition is less intense. For example, among coffee brands, niche coffee brands with strong design sense like Manner, M Stand, and Blue Bottle can command higher prices. Similarly, well - known watch brands like Rolex can sell at a higher premium, and Patek Philippe, with the slogan "You never actually own a Patek Philippe. You merely look after it for the next generation", has an even higher premium than Rolex.

So far, we haven't seen significant large - scale applications of Generative AI in terms of emotional, life - changing, and social impact user values. Although there have been some initial successful cases of Generative AI in applications such as emotion - related, chat, and gaming, most of them are in niche or sub - cultural fields, and a "killer - app" with mass appeal like WeChat has not emerged yet.

Therefore, when formulating strategies in the AI era, enterprises should first focus on how to create value for users, especially emotional, life - changing, and social impact values beyond functional value, and think about and explore how to use Generative AI to create these types of user values.

Focus on Unique Value Contribution

According to the value - based strategic analysis framework, among the value created by an enterprise through cooperation with customers and suppliers, only a part (the "price - cost" part, i.e., enterprise profit) goes to the enterprise, while the rest (the "customer's willingness to pay - price", i.e., consumer surplus, and the "cost - supplier's opportunity cost", i.e., supplier surplus) is distributed to customers and suppliers respectively.

The amount of the jointly created value that an enterprise, customers, and suppliers can each get depends on the size of their unique contributions in the value - creation process, that is, "the total value created with the company's participation - the total value created without the company's participation". This difference is also called "Added Value", which can also be understood as "the additional value that the enterprise can bring". The larger this difference is, the greater the enterprise's say in the distribution of the created value and the larger the share it can get.

Understanding this, we can also understand why the market values of core enterprises such as NVIDIA, TSMC, and ASML in the AI industry chain have been continuously hitting new highs in the past three years. Because they have made indispensable and almost unique key contributions in the rapidly developing AI industry chain. In the entire industry chain, the "additional value that they can bring" is large enough and keeps growing rapidly, bringing them substantial and rapidly growing profits and supporting the rapid growth of their market values.

Similarly, in the past three years, the market values of enterprises in related sectors such as liquid cooling, power (nuclear power), and storage in the AI industry chain have also increased significantly, and many have even exceeded the market - value growth of core enterprises like NVIDIA.

A recent example is that the launch of Google's Gemini 3 and TPU technology has made the market worried about the "uniqueness" and "indispensability" of OpenAI in large models and NVIDIA in the GPU business, which has led to a recent (this week) decline in NVIDIA's stock price while Google's stock price hit a new high.

Correspondingly, in the downstream (application end) of the AI industry chain, there haven't been particularly outstanding applications or enterprises in the past three years (except for a few enterprises like Palantir and Applovin). The reason is that there haven't been "killer - apps"/"mass - apps" like Amazon, Taobao, or WeChat that are relatively "unique" and "hard to imitate/substitute" in the downstream application end.

Interestingly, large enterprises like Tencent, Alibaba, and ByteDance have used the power of Generative AI to quickly follow up and amplify their existing advantages, and their market values or valuations have increased significantly.

Therefore, when formulating strategies in the AI era, in addition to focusing on user value creation, enterprises also need to think about and explore how to make or play a relatively unique contribution in the value - creation process so as to have a relatively strong say and the right to distribute benefits in the rapidly growing AI era.

Explore and Build a Moat with "Economies of Scale + Network Effects"

Strategy is about the future and is forward - looking. Therefore, in addition to user value creation and unique value contribution, enterprises should also think about where and what their future moats will be.

Classic enterprise and business moats such as brand, patent/proprietary technology, location, and switching cost are well - known. But from a dynamic and long - term perspective, enterprises need to focus more on whether their products and services can establish and achieve economies of scale or network effects.

From the perspective of modern business history, great enterprises usually can achieve economies of scale, network effects, or both (such as Maotai, Tencent, and Amazon). In the wave of Generative AI in the past three years, NVIDIA firmly established a multi - sided network effect connecting enterprise users and development users through the CUDA development environment it started to layout 20 years ago, thus building a strong moat for itself in the fierce AI computing - power competition.

Therefore, when thinking about and formulating strategies in the AI era, we still need to be forward - looking, carefully think about and plan how to continuously achieve and improve economies of scale or network effects, or achieve both in the future.

Summary and Outlook

At this moment, there are still some uncertain issues in the development prospects of technologies represented by Generative AI, including but not limited to: 1) Whether Generative AI/large language models can break through the basic principle of probability statistics and thus have causal reasoning ability, getting closer to the general artificial intelligence that everyone expects; 2) Whether the physical AI model can make a breakthrough to achieve a breakthrough in the direction of embodied intelligence related to hardware; 3) How to solve the resulting data - security and information - source pollution problems; 4) In the next 1 - 2 years, whether the investments in chips and computing power by the upstream and downstream of the AI industry in the past three years can be verified and supported by the revenues of the downstream application end, etc.

But no matter what the future holds, when formulating strategies, we must fully focus on user value creation, focus on unique value contribution, and explore and build our own moats as early as possible. Only in this way can we build an excellent corporate strategy in the AI era and establish and maintain a competitive advantage.

Professor Introduction

Dr. Zhang Yu is a professor of strategy at the China Europe International Business School (CEIBS), the director of the Department of Strategy and Entrepreneurship, and the co - director of the CEIBS Research Center for Innovation and Entrepreneurship. Professor Zhang obtained his Ph.D. in management from INSEAD. Before joining CEIBS, he was an assistant professor of strategy at the Paul Merage School of Business at the University of California, Irvine. His teaching areas mainly include strategic management, industry and competition analysis, innovation and competition, corporate transformation and upgrading, etc. Professor Zhang's research interests mainly focus on the interaction between strategy and capital markets. His research has been published in top international journals such as The Academy of Management Journal, Organization Science, and Strategic Management Journal. He has been invited to give academic speeches at well - known business schools in Europe, America, and Asia and keynote speeches at well - known industry conferences such as the annual meeting of the National Association of Corporate Directors.

Author's Note: This article was written without using any AI tools. It is based on the known information as of November 27, 2025, and also benefits from discussions and exchanges with CEIBS students and alumni in and out of the classroom and at sharing events in the past three years. As of November 27, 2025, except for Palantir, I do not hold long or short positions in the stocks of other enterprises mentioned in this article.

① Carl Shapiro, Hal R. Varian. (2000). Information Rules: A Strategic Guide to the Network Economy, First Edition in 2000, Beijing, China Renmin University Press

② Brandenburger, A. M., & Stuart Jr, H. W. (1996). Value‐based business strategy. Journal of economics & management strategy, 5(1), 5 - 24.

③ Discovering Customers' Real Needs with Value Elements, Harvard Business Review Chinese Edition, September 2016

④ For 2B businesses, refer to The B2B Value Pyramid, Harvard Business Review Chinese Edition, May 2018

This article is from the WeChat official account "China Europe International Business School" (ID: CEIBS6688), author: Zhang Yu, published by 36Kr with authorization.