Lobster Moment and DeepSeek Impact Kimi K3 has received widespread praise across the internet, with everyone saying it will benchmark against Fable5; The rise of Chinese big models has raised questions about the pricing of OpenAI and Anthropic. Deepseek shocks are bound to occur every once in a while, and as long as the US big model factories don't have anything new to widen the gap for a few months, Chinese big models will drag their American competitors into a price war at 1/20 of the price and 90% of the ability. So from the perspective of investment, it is also a gradual back and forth relationship: the US big model is impacted, the high investment in semiconductor Capex is difficult to sustain, and compared with Chinese big models and domestic computing power chains, the cost-effectiveness of US investment is questioned, which is the Deepseek moment. Companies with less Capex investment and strong cash flow, such as Apple and Tencent, will see an increase, while companies with more Capex investment and tight cash flow, such as Amazon, will see a decrease. The semiconductor industry chain (storage, CPU, optical modules, etc.) will also decline; A few months later, the United States continued to launch new products to widen the gap, the semiconductor super cycle continued to be launched, and cloud factories continued to invest; Companies with high investment and the semiconductor industry chain continue to rise, similar to the lobster outbreak at the beginning of the year with Opus 4.6; The intelligent economy of agents has ignited the demand for tokens. The development of AI is a tortuous journey, with DeepSeek and Lobster moments alternating every six months. This mode has two endings: The ultimate goal of health is for the US big model to continue to make efforts to widen the gap, cloud factory revenue ROI to continue to improve, cash flow to continue investing, semiconductor demand to continue to increase, and supply and demand to be strong. Hello everyone, continue to dance and continue to be high; The tragic ending is that the American big model can no longer go up, and each update is incremental rather than jumping; Analogous to the iPhone being updated every year, but after the iPhone 4, there gradually disappeared the revolutionary gap, squeezing toothpaste every year; The cloud factory has invested heavily, but its cash flow still cannot turn positive, and its ROI continues to decline. The price war of China's big model knocked down the price of the big model, and the first level valuation of the two OA plants began to hang upside down. The big model fell into the price war, and the asset foam collapsed. After the price war, the cost of the big model decreased, the penetration rate increased, and finally realized positive return on capital, and walked out of the foam to replicate the path of Internet recovery. The current pricing of assets is moving towards a pessimistic direction. To return to an optimistic situation, we can only hope that the OA two factories will release big moves, and that the large model capability will have a leapfrog development, driving the demand for large models to continue to increase. The cloud factory will restore balance by leasing computing power. We look forward to the next lobster moment