Under the surface glory, OpenAI's' four major challenges' Author: Zhao Ying Source: Wall Street Insights Benedict Evans, former partner and renowned technology analyst of a16z, recently published an in-depth analysis article, pointing out that OpenAI is facing four fundamental strategic dilemmas behind its apparent prosperity. He believes that although OpenAI has a large user base and sufficient capital, the lack of a technological moat, insufficient user stickiness, rapid catch-up from competitors, and product strategy constrained by laboratory research and development directions are threatening its long-term competitiveness. Evans pointed out that OpenAI's current business model does not have a clear competitive advantage. The company has neither unique technology nor network effects, with only 5% of its 900 million weekly active users paying, and 80% of users sending less than 1000 messages by 2025- equivalent to an average of less than three prompts per day. This usage pattern of 'one mile wide, one inch deep' indicates that ChatGPT has not yet become a daily habit for users. Meanwhile, tech giants such as Google and Meta have caught up with OpenAI in terms of technology and are leveraging their distribution advantages to seize market share. Evans believes that the true value of the AI field will come from new experiences and application scenarios that have not yet been invented, and OpenAI cannot create all of these innovations alone. This requires the company to operate on multiple fronts simultaneously, with a comprehensive layout from infrastructure to application layer. Evans' analysis reveals a core contradiction: OpenAI attempts to establish competitive barriers through large-scale capital investment and a full stack platform strategy, but there are still doubts about whether this strategy can be effective in the absence of network effects and user lock-in mechanisms. For investors, this means reassessing OpenAI's long-term value proposition and its true position in the AI competitive landscape. The disappearance of technological advantages: The homogenization of models has intensified. Evans pointed out in his analysis that there are currently about six institutions that can launch competitive cutting-edge models with similar performance. Every few weeks, companies surpass each other, but no one can establish a technological leadership position that other companies cannot match. This is in stark contrast to platforms such as Windows, Google Search, or Instagram, which have self reinforced their market share through network effects, making it difficult for competitors to break their monopoly no matter how much money and effort they invest. The situation of technological equality may change due to certain breakthroughs, most notably the implementation of continuous learning capabilities, but Evans believes that OpenAI is currently unable to plan for this. Another possible differentiating factor is the scale effect of proprietary data, including user data or vertical industry data, but existing platform companies also have advantages in this regard. In the context of model performance convergence, competition is shifting towards brands and distribution channels. The rapid growth in market share of Gemini and Meta AI confirms this trend - for ordinary users, these products may seem similar, while Google and Meta have strong distribution capabilities. In contrast, Anthropic's Claude model, although often ranked high in benchmark tests, has close to zero consumer awareness due to a lack of consumer strategy and product. Evans compared ChatGPT to Netscape, which had an early advantage in the browser market but was ultimately defeated by Microsoft using its distribution advantage. He believes that chatbots and browsers face the same differentiation challenge: they are essentially just input and output boxes, and the space for product innovation is extremely limited. Despite OpenAI's clear lead of 800 to 900 million weekly active users, Evans pointed out that this data masks a serious issue of user engagement. The vast majority of users who already understand and know how to use ChatGPT have not developed it into a daily habit. Data shows that only 5% of ChatGPT users pay, and even among American teenagers, the proportion of using it a few times a week or less is much higher than the proportion of using it multiple times a day. OpenAI revealed in its "2025 Annual Summary" campaign that 80% of users will send less than 1000 messages in 2025, which is equivalent to an average of less than three prompts per day based on face value, and even fewer actual chats. This shallow usage means that most users cannot see the differences in personality and emphasis between different models, nor can they benefit from features such as' memory 'aimed at building stickiness. Evans emphasized that memory function can only bring stickiness, not network effects. Meanwhile, usage data from a larger user base may be an advantage, but the extent of this advantage is questionable when 80% of users only use it a few times a week at most. OpenAI itself acknowledges the existence of problems, stating that there is a 'capability gap' between the model's capabilities and the actual usage by users. Evans believes that this is avoiding the fact that the product market fit is unclear. If users can't think of what to do with it on ordinary days, it means it hasn't changed their lives yet. The company has launched an advertising project, partly to cover the service costs of over 90% of non paying users, but more strategically, it enables the company to provide these users with the latest, most powerful (and expensive) models, hoping to deepen user engagement. However, Evans questioned whether providing users with better models could change this situation if they couldn't figure out what to do with ChatGPT today or this week. Doubts about Platform Strategy: Lack of Real Flywheel Effect Last year, OpenAI CEO Sam Altman attempted to integrate the company's various initiatives into a coherent strategy, presenting a chart and quoting Bill Gates' famous quote: The definition of a platform is creating more value for partners than for itself. At the same time, the Chief Financial Officer released another chart showcasing the 'flywheel effect'. Evans believes that the flywheel effect is a sophisticated and coherent strategy: capital expenditures themselves form a virtuous cycle and become the foundation for building full stack platform companies on top of it. Starting from chips and infrastructure, build each layer of technology stack upwards. The higher you go, the more you can help others use your tools to create their own products. Everyone uses your cloud, chips, and models, and then at higher levels, the layers of the technology stack reinforce each other, forming network effects and ecosystems. However, Evans bluntly stated that he believes this is not the correct analogy. OpenAI does not possess the platform and ecosystem dynamics that Microsoft or Apple once had, and the flywheel diagram does not actually show the true flywheel effect. In terms of capital expenditures, the four major cloud computing companies invested approximately $400 billion in infrastructure last year and announced at least $650 billion this year. OpenAI claimed a few months ago that it has a promise of $1.4 trillion and 30 gigawatts of computing power in the future (with no clear timeline), but the actual usage by the end of 2025 is 1.9 gigawatts. Due to the lack of large-scale cash flow from existing businesses, the company achieves these goals through financing and the use of others' balance sheets (partially involving 'revolving income'). Evans believes that large-scale capital investment may only gain a seat, rather than a competitive advantage. He compared the cost of AI infrastructure to that of aircraft manufacturing or the semiconductor industry: without network effects, the processes of each generation of products become more difficult and expensive, and ultimately only a few companies can sustain the investment needed to stay at the forefront. However, although TSMC has a de facto monopoly position in the cutting-edge chip field, this has not brought it leverage or value acquisition capabilities in the upstream technology stack. Evans pointed out that developers must build applications for Windows because it has almost all users, and users must purchase Windows PCs because it has almost all developers - this is the network effect. But if you use generative AI to invent an excellent new application or product, you only need to call the underlying model running in the cloud through APIs, and users do not know or care about what model you are using. Lack of product dominance: Strategy constrained by laboratory Evans quoted a passage from OpenAI product leader Fidji Simo in 2026 at the beginning of the article: 'Jakub and Mark set long-term research directions.'. After months of work, amazing results emerged, and researchers would contact me saying, 'I have some cool stuff. How do you plan to use it in chat? How will it be used for our enterprise products?' This statement contrasts sharply with Steve Jobs' famous quote in 1997: 'You must start with customer experience and then reverse engineer to technology. You cannot start with technology and then try to figure out where to sell it.' Evans believes that when you are a product leader in an AI lab, you cannot control your roadmap and your ability to set product strategies is very limited. You opened your email in the morning and found that the laboratory had researched something, and your job was to turn it into a button. Strategy happens elsewhere, but where is it? This issue highlights the fundamental challenge that OpenAI faces: unlike Google in the 2000s or Apple in the 2010s, OpenAI's smart and ambitious employees do not have a truly effective product that others cannot achieve. Evans believes that one interpretation of OpenAI's activities over the past 12 months is that Sam Altman is deeply aware of this and is attempting to translate the company's valuation into a more sustainable strategic position before the music stops. For most of last year, OpenAI's answer seemed to be 'everything, done simultaneously, executed immediately'. Application platforms, browsers, social video applications, collaborations with Jony Ive, medical research, advertising, and more. Evans believes that some of them may seem like a 'comprehensive attack' or simply the result of quickly recruiting a large number of proactive individuals. Sometimes it gives people a feeling that they are copying the form of previous successful platforms without fully understanding their purpose or dynamic mechanisms. Evans repeatedly uses terms such as platform, ecosystem, leverage, and network effects, but he acknowledges that these terms are widely used in the technology industry and their meanings are quite vague. He quoted Roger Lovatt, a medieval history professor from his university days, as saying: Power is the ability to make people do things they don't want to do. This is the real question: Does OpenAI have the ability to enable consumers, developers, and businesses to use its system more, regardless of what the system itself actually does? Microsoft, Apple, and Facebook used to have this ability, and Amazon did the same. Evans believes that a good way to interpret Bill Gates' statement is that the platform truly leverages the creativity of the entire technology industry, so that you don't have to invent everything yourself and can build more things on a large scale, but all of this is done on your system and controlled by you. The basic model is indeed a multiplier, and a large number of new things will be built using them. But do you have a reason for everyone to have to use your product, even if competitors have already built the same thing? Is there a reason for your product to always outperform competitors, regardless of how much capital and effort they invest? Evans concluded that without these advantages, the only thing you have is daily execution. Executing better than others is certainly a wish, and some companies have achieved this over a longer period of time, even convincing themselves to institutionalize it, but this is not a strategy.