History may be repeating itself if China transforms AI into a widely available, commodity-driven industry. Should we be concerned about AI becoming a new kind of abundant resource? When a Beijing-based AI firm, Zhipu, released an open-source General Language Model (GLM) that began competing with Anthropic’s Claude and OpenAI’s ChatGPT at a much lower price, a familiar question arose: are we witnessing the birth of a new kind of intelligence, or a familiar pattern from the past? Western companies like Microsoft and NVIDIA have long relied on the idea that AI models are scarce, expensive, and controlled by those who train them. This scarcity has been a key part of their financial success, allowing them to maintain a competitive edge in the market. However, if a powerful AI model can be freely downloaded and used, the value of owning one diminishes significantly. This is a scenario that Europe has seen before, when it developed the solar industry but later watched its manufacturers disappear as Chinese companies flooded the market with cheaper alternatives. China's approach to AI is not new. It has followed a pattern seen in other industries where Western innovation was first developed and then replicated on a massive scale by Chinese firms. In these cases, state-owned banks funded the necessary factories, and the resulting products were sold at prices that undercut Western competitors. This strategy, while financially unappealing for private investors, serves political goals such as employment, prestige, and strategic dominance. When state-backed companies can afford to operate at a loss, they can set prices that make it difficult for private firms to compete. The impact of this approach is clear. If AI models become abundant and cheap, the financial advantage of the original innovators is reduced. This could challenge the upcoming initial public offerings (IPOs) of companies like Anthropic and OpenAI, which are banking on improving their operating profits. Meanwhile, the adoption of AI technology is also influenced by cost. While American labs may lead in elite benchmarks, the systems that end up being widely used are often the ones that are good enough and inexpensive. In contrast to AI, technologies like solar panels and electric vehicles have evolved gradually, with improvements in efficiency and cost effectiveness driving competition. These technologies follow what is known as Moore’s Law, which suggests that computing power doubles roughly every two years while costs fall. Similar trends are seen in other areas, leading to increased competition and the eventual commodification of these products. AI models, however, are different. They are trained once and then apply what they have learned. Unlike the human brain, which continuously adapts, current AI models require retraining with more data, more computing power, and more energy to become more capable. This means they are closed systems, limited in their abilities once training ends. If AI models were to evolve continuously, like the human brain, the landscape would change dramatically. Such systems would compound their advantages rather than compete for resources. Until that happens, the competitive edge of AI models will likely erode, much like it has for other technologies. The value will shift from owning an AI model to building applications and systems that use AI in innovative ways. China is already leading in this direction, showing that the real value lies in how AI is used, not just how "smart" the models are.