The United States has started to regulate AI, and the valuation logic of the AI bull market is changing ----From computing power, models to national security, AI has been labeled as a "regulatory discount" for the first time. The incident that occurred on June 13th really shook the AI field and financial markets: The US government directly requested Anthropic to implement export controls on its state-of-the-art Fable 5 and Mythos 5 models on the grounds of national security. According to media reports, this restriction not only applies to overseas users, but also includes foreign citizens within the United States. Due to the difficulty in accurately identifying the visitor's identity, Anthropic ultimately chose to directly shut down the relevant model services. The importance of this matter far exceeds that of a company's product being taken offline. Because this is the first time that the United States has publicly managed top-level large models as strategic resources similar to advanced chips and dual-use technologies. In the past, the United States restricted the tools for manufacturing AI, but now it is beginning to limit the diffusion of AI capabilities themselves. I am not surprised at all that the US government regulates and restricts AI big models. After all, in the past 10 years, the US government's technology war, financial war, trade war, and talent war against China have all demonstrated its fear of China's rapid technological development. It has had to adopt various means to restrict and contain China in multiple dimensions, channels, and fronts. Today it is the turn of the US government to restrict the export of AI models, which is just a regulatory action taken to maintain the last bit of pride in capital and technology. In the past two years, the AI market has almost had only one core narrative: whoever has stronger models and more computing power will receive higher valuations. From NVIDIA to Microsoft, from OpenAI to Anthropic, the valuation logic of the entire industry chain revolves around the same formula: the stronger the model capability, the faster the user growth, the greater the future commercialization space, and the higher the premium that capital is willing to offer. But a series of recent events are quietly changing this set of game rules. The US government has begun to strengthen the management of top AI models, and access permissions for some of the most advanced models have been included in the national security framework for review. This means that for the first time, the AI industry is no longer just a simple technology industry, but has begun to possess the attributes of strategic assets. In the past two years, the United States has mainly restricted the export of AI chips and advanced computing power, such as high-end GPUs like H100. After entering 2026, regulation will further extend from "computing power" to "models" themselves. In early June, the White House issued a new AI security executive order proposing the establishment of a Frontier Models national security assessment system; Subsequently, the United States requested Anthropic to implement access restrictions on some of its most advanced models on the grounds of national security. This sends a very clear signal: the United States is starting to view top AI models as strategic resources, rather than just ordinary commercial products. For the market, this is not just an ordinary regulatory news, but a completely new valuation variable is emerging. In the past, when investors evaluated an AI company, they mainly focused on model capabilities, computing power reserves, user growth, and commercialization progress. In the future, national security, policy risks, export restrictions, and international access may also enter valuation models. Simply put, the AI industry is beginning to experience a 'regulatory discount'. This kind of change is not unfamiliar. The semiconductor industry has gone through a similar process. When the United States began to restrict the export of advanced chips, the market gradually realized that chips are not just commodities, but also strategic resources. Nowadays, the same logic is extending to the AI model itself. If the United States has mainly restricted GPUs, advanced processes, and computing infrastructure in the past two years, then what is now being touched upon is the model itself. This is an important watershed. Because chip limitations target production capacity, while model limitations target final capability. When a country begins to restrict access and dissemination of the most advanced AI models, it is essentially managing future productivity tools. This is also why more and more investment institutions are reassessing the long-term value of AI companies. In the past, the market assumed that excellent models could be sold globally, and global developers could become potential customers. But if the most advanced models in the future need to undergo export reviews, access restrictions, or even national security assessments, then the market space that was originally taken for granted may not be fully realized. For the capital market, expected changes are often more important than reality. Even if the restrictive measures have not been fully implemented, a change in regulatory direction alone is enough to make investors recalculate the risk return ratio. At the same time, another noteworthy trend is gradually taking shape. In the past few years, although the competition in the AI industry between China and the United States has been fierce, they are still essentially in the same technological ecosystem. The United States is responsible for leading the development of cutting-edge models, while China is rapidly catching up and commercializing them. Although there are limitations between the two parties, they still maintain a certain degree of technical connection. But now, this state is undergoing a change. The United States is beginning to view the most advanced AI models as strategic resources, while China is accelerating the construction of its own model system and computing power system. From DeepSeek to Tongyi Qianwen, from Doubao to Hunyuan, Chinese companies are establishing their own AI ecosystem. In the coming years, the global AI industry is likely to no longer be a unified market, but gradually evolve into two relatively independent technological systems. This is a bit like the Internet 20 years ago. The United States has Google, Amazon, and Meta, while China has Baidu, Alibaba, and Tencent. Both parties share similar technological routes, but have formed their own independent ecosystems. And today's big model industry seems to be heading towards a similar path. Of course, this does not mean that the AI bull market is over. In fact, there may be two completely different interpretations of the market in the future. Pessimists believe that regulation means restrictions, and restrictions mean shrinking commercialization space, so the valuation of AI companies needs to be lowered again. Optimists, on the other hand, believe that when a technology is incorporated into the national security framework, it precisely indicates that its importance is on the rise. Regulation may not necessarily weaken the position of leading companies, but may further raise industry barriers and provide deeper moats for top enterprises. From this perspective, regulation can be both a discount and a premium. The key is how the market will ultimately understand it. More noteworthy is that all of this happened just before and after SpaceX's launch. In history, super IPOs often meant that market risk appetite was approaching a phase of climax. When investors are willing to pay increasingly high prices for the hottest technology assets, any new uncertainty could become a catalyst for repricing. Therefore, what is truly worth observing next is not whether the United States will continue to strengthen regulation, but how the market will respond to this change. Will investors view AI as a constrained industry or an industry endorsed by national strategies? These are two completely different stories and will correspond to two completely different valuation systems. It can be confirmed that the AI industry is entering a new stage. In the past, the market traded on models, computing power, and growth; In the future, the market will also have trading policies, regulations, and national security. When these variables begin to enter the pricing system, the logic of the AI bull market will also change accordingly. And this may be one of the most noteworthy investment themes for the second half of 2026. The AI competition between China and the United States will become increasingly complex and fierce, and investing in the AI industry also requires choosing sides.