A New Map of Supply Chains, Regulations, and Capital: Insights from Huawei, AMAT, BlackRock, and Perplexity
This article traverses the latest topics in semiconductors, generative AI, EVs, data centers, and media, re-examining them through the four quadrants of "supply chain," "regulation," "capital," and "demand." Based on program statements and news summaries, it organizes a cross-industry overview including Huawei, TSMC, Samsung, NVIDIA, Applied Materials, Rivian, Tesla, Perplexity, OpenAI, BlackRock GIP, Aligned Data Centers, Apple, and even Taylor Swift, who symbolizes entertainment demand.
1. Semiconductors: Sanction "loopholes" and the wall of self-sufficiency
HuaweiAI processor teardowns have pointed to the presence of HBM "2E" generation components from companies like Samsung. While the procurement route under sanctions remains opaque, it highlights the reality that "domestic supply in China cannot cover everything." While the Chinese government is pushing for domestic purchases from the demand side, if access to advanced nodes (3nm/GAA) remains blocked, the risk of a "plateau" in production increases. The desire to find alternatives to NVIDIA remains strong, but component constraints will dictate the sustainability of competitiveness.
1-1. Spillover to manufacturing equipment
Applied Materials mentioned a sales impact of approximately $600 million due to tightened regulations on China. As the saying goes, "chips touch AMAT equipment at the beginning and the end," the equipment sector is highly susceptible to direct hits from regulations. Equipment and materials are the peripheral industries most sensitive to geopolitical risk.
2. Regulation x Diplomacy: The condition for trade is "technology"
Reports have suggested moves by China to explore easing China-related regulations stemming from national security in exchange for expanding investment in the U.S. (on a trillion-dollar scale). The technical constraint that "you cannot advance to next-generation AI architecture without reaching 3nm/GAA" is at the heart of the negotiations. Investment capital is the "gasoline," but the key is whether technology transfer is possible.
3. Infrastructure: The power and capital sucked up by AI
BlackRock GIP is reportedly in the final stages of a Aligned Data Centers acquisition worth approximately $40 billion. While not flashy, the bottlenecks for AI demand are power, cooling, and land, making data centers the "underground infrastructure" of the AI value chain. This also suggests the possibility that M&A is shifting to a "buy rather than build" phase.
4. Software x Browser: Reorganization of the AI-native experience
Perplexity has released its AI-native browser (Comet) for free. Behavioral changes have been observed, with "users asking 6 to 18 times more questions." For monetization, they are proposing a model that includes Pro/Max subscriptions and distribution to publishers via Comet Plus. They are challenging the existing web economy, which is skewed toward search and advertising, with an alternative of "question-driven x flat-rate access." Competition for the winner of distribution, including developments for Apple, is accelerating.
5. Generative Video: The tension between convenience and risk
OpenAI has launched its latest video generation app. While they are implementing measures such as watermarks, demonstrations showing that "it could be bypassed by screen recording at the draft stage" indicate that the asymmetry of ease of diffusion vs. detection remains. Regulation, disclosure, and the standardization of watermarks are urgent tasks to be developed as "cross-industry public goods."
6. Mobility: UI/UX for safety is the "emergency standard"
Rivian is revising the manual door release for the R2 to be in a more intuitive position in the event of power loss. Similar modifications have also become a topic of discussion at Tesla. Beyond aesthetics and cost in normal times, primary operability in emergencies is becoming a new design standard. "Ease of discovery" and "consistency of procedures (front and rear seats)" will likely also impact regulations and insurance premiums.
7. Market: The question of AI "overheating" and the art of diversification
"Valuations are high, but discern the new normal of scale."—The guest's assessment hits the nail on the head. If the scope expands beyond NVIDIA and OpenAI to peripheral industries like construction machinery (Caterpillar), the cost of concentration—individual risk—will increase. Recommended actions include profit-taking, reallocation, and picking up adjacencies. As with the bearish assessment of Apple, one must also be mindful of the "time lag in AI benefits."
8. The rumble of demand: "Simultaneous connection" shown by culture
Taylor Swift's new album has dominated search trends. With the movie release overlapping, cross-platform customer acquisition has been achieved. The killer app in the generative AI era is, after all, "places where people gather," which triggers a chain reaction from CDN/DC/ad inventory/payment funnels to boost demand.
Summary
The big picture is clear. Technical access (3nm/GAA) and regulations determine the ceiling for semiconductors, capital is targeting data centers, UI/UX and safety are redesigning EVs, and AI-native distribution infrastructure is rebuilding the web economy. For investors, there are three practical tasks: (1) Managing volatility in equipment and materials with high sensitivity to geopolitical risk. (2) Focusing on data center assets with strong power and location advantages, along with their surrounding cash-generating capabilities. (3) Understanding the 'question-driven' distribution and revenue-sharing models of the creator economy. While debates over overheating continue, demand ultimately boils down to physical infrastructure and institutional design. That is where the source of long-term alpha lies.
