SemiAnalysis, a research firm that started with a personal blog and can now crash the US stock market with just one report, and its founder @ dylan522p, who has a non semiconductor background: 1. Dylan Patel was born in 1996 and holds a Bachelor's degree in Management and Law from the University of Georgia in the United States. During adolescence (8-12 years old), active in online hardware forums, as a 'forum warrior', self-learning Xbox and other hardware maintenance and chip technology, mastering professional knowledge through reading documents and community communication. For many years, I have anonymously published chip and hardware analysis content on Reddit, WordPress blogs, and Silicon Twitter, as well as moderated multiple Nvidia and other hardware communities. 2. On May 22, 2020, Dylan posted his first blog on his 24th birthday and officially founded SemiAnalysis thereafter. Initially started as a personal anonymous WordPress blog, later transitioned to Substack paid content. In the early days, Dylan focused solely on semiconductor supply chain, AI infrastructure, cloud ecosystems, machine learning models, and deep analysis in related fields, gradually developing into a consulting business. Dylan himself has grown from an anonymous "Silicon Twitter chip blogger" to one of the most followed AI infrastructure analysts worldwide. 3. In recent years, SemiAnalysis has rapidly expanded and has now developed into the world's leading AI infrastructure and semiconductor research institution, with a team size of about 60 people distributed in 8-10 countries/regions including the United States, Japan, Taiwan, Singapore, France, Germany, Israel, Canada, and the United Kingdom. SemiAnalysis is expected to generate over $100 million in revenue in 2026 (The Information reports), achieving explosive growth compared to the early subscription model. The business model includes: 1) Paid subscription (newsletter+in-depth reports); 2) Sales of proprietary industry models (AI Datacenter Model, Accelerator Industry Model, Wafer Fab Model, AI Cloud TCO Model, AI Networking Model, etc., used to predict wafer production capacity, data center power, accelerator output, TCO economics, etc.); 3) Consulting services, customized projects, and hourly consultations. Provide a unified perspective for the entire supply chain (wafer fab → cloud → ML model). These proprietary models and data (such as satellite image monitoring data center construction, GitHub commit analysis, HBM supply chain tracking) have become core decision-making tools for customers, including top AI labs, hyperscalers, hedge funds, and semiconductor giants. Their analysis directly affects trillion dollar AI capital expenditure decisions. 4. More importantly, SemiAnalysis's current super influence 1) At this year's GTC keynote, Huang Renxun publicly named Dylan Patel and presented SemiAnalysis's latest InferenceX chip performance report on the big screen. He spent a full 5 minutes interpreting it and continued to cite it in Q&A. Jensen, a big shot at this level, proactively called out an independent analyst, and the value was directly maximized. 2) Su Zifeng is even more ruthless: SemiAnalysis released a report criticizing the training performance of AMD MI300X. The next day, Lisa Su personally called to arrange a 90 minute one-on-one talk, and later publicly tweeted to thank "feedback is a gift even when it's critical". This operation is like pulling an independent research institution directly into a CEO level strategic dialogue. From this perspective, SemiAnalysis is no longer just a "newsletter", but a strategic intelligence hub for the AI semiconductor ecosystem. Its influence is amplified through proprietary data models, high-frequency media exposure, and bilateral customer service, directly serving engineering decision-making in Silicon Valley and billion dollar transactions on Wall Street. Nowadays, when hedge funds and asset management institutions make AI/semiconductor positions, SemiAnalysis's monthly ChipBook and proprietary data are already standard configurations - idea generation, thesis validation, and portfolio tracking all rely on it. Once the report is released, it often directly drives stock price fluctuations and billions of capex adjustments. Hyperscalers planning custom ASICs, GPU procurement, and data center power all depend on their supply chain breakdown. It is no longer a 'reference', but a direct influence on decision-making. This is a typical "non-traditional 0-1" approach: without an EE PhD from a prestigious university or a background in a major company, one relies on self-learning, data-driven approaches, and independent thinking, from working on anonymous blogs to taking NVIDIA and AMD CEOs seriously. Silicon Valley and Wall Street now default - to figure out the true rhythm of AI hardware, first look at what SemiAnalysis and Daylan have said. This is probably the state where a person can achieve the ultimate level of independent research and investment.