Indepth Research

Provide in-depth research reports and independent analysis, leveraging data, technology, and economic insights to deliver a comprehensive examination of the blockchain ecosystem, project potential, and market trends.

Why the Establishment of SocialFi Originates from a Misunderstanding of Its Own Medium

"Why SocialFi's Establishment Stems from a Misunderstanding of Its Own Medium" This article critiques the failure of SocialFi projects by applying Marshall McLuhan's theory of "hot" and "cool" media. McLuhan posited that a medium's form—not its content—reshapes user behavior. "Hot" media (e.g., print, radio) deliver high-definition, complete information, promoting passive consumption. "Cool" media (e.g., cartoons, telephone calls) provide low-definition, fragmented signals, requiring active user participation to complete the meaning. Traditional social media platforms (like early Twitter) are quintessentially "cool." A tweet or like is an incomplete fragment; its significance emerges only through replies, shares, and community engagement—it's a participation engine disguised as a content system. SocialFi (e.g., Friend.tech) aimed to monetize social capital by attaching real-time, tradable prices to follows and posts. However, this didn't add an economic layer to a cool medium; it fundamentally transformed the medium itself. The explicit, high-resolution signal of price replaced the ambiguous, low-resolution signal of social interaction. The platform became a financial market dressed as a social network. Once the financial dynamics (speculative profits) faded, the underlying social fabric, which had been suffocated from the start, could not sustain it. The medium overheated and collapsed. This "heat death" pattern isn't unique to crypto. Over time, mainstream platforms often drift from cool to hot by adding features like public metrics, verification badges, and algorithmic feeds that optimize for clarity over participation, leading to user disengagement. The article proposes a viable alternative: the "condensation point." Here, capital is introduced locally and infrequently into a cool medium without saturating it. Examples include Substack (subscriptions), Patreon (memberships), and Bandcamp (music purchases). The core social medium remains cool and participatory, while capital condenses at specific, structurally separate points (e.g., a monthly fee). The key lesson: "Liquidity is heat." Adding it to a cool medium doesn't enhance it but alters its fundamental nature. The NFT boom and bust provides a starker example. Collecting is a classic cool medium, where value is built slowly through stories and community. By making floor prices, rarity scores, and real-time charts omnipresent, NFT platforms rapidly overheated the medium, turning collectors into traders and destroying the participatory culture that gave collections meaning in the first place. The conclusion is that for the next wave to succeed, designers must ask not how to price every social action, but how to let capital condense within a social system without disrupting the cool, participatory mechanics that create its enduring value.

marsbit05/14 09:39

Why the Establishment of SocialFi Originates from a Misunderstanding of Its Own Medium

marsbit05/14 09:39

After Storage, Are Copper and Fiber Optic Cables Facing an AI "Great Famine"?

Following the storage sector, copper and fiber optics are emerging as potentially the next major markets to experience explosive growth due to AI. Demand for copper, described by Goldman Sachs as "the oil of the AI era," is surging. Prices are near record highs, with LME copper up 41% over the past 12 months. This is driven by AI's immense and unique requirements: copper is the essential material for the massive electrical distribution (e.g., a 1GW AI data center requires ~27,000 tons) and advanced liquid cooling systems needed for high-power AI clusters like NVIDIA's GB200. Meanwhile, new large-scale copper mine discoveries have been scarce for a decade, tightening supply. Concurrently, a "fiber famine" is unfolding. AI's need for ultra-high-speed, long-distance interconnects between thousands of GPUs is pushing data transmission beyond the physical limits of copper cables. Demand for fiber optics is experiencing a step-change, with a single AI data center requiring up to 36 times more fiber than a traditional CPU rack. This has caused prices for standard G.652D fiber in China to nearly double in just three months. Supply is critically constrained due to the long (18-24 month) lead times required to expand production of the core preform material. In summary, AI's infrastructure demands are cascading down from semiconductors to foundational materials. Copper faces a structural supply-demand imbalance, while fiber optics is entering a period of severe shortage, positioning both as critical and potentially strained components of the AI build-out.

marsbit05/14 09:25

After Storage, Are Copper and Fiber Optic Cables Facing an AI "Great Famine"?

marsbit05/14 09:25

The Construction of SocialFi Originates from a Misreading of Its Own Medium

This article argues that the fundamental failure of SocialFi projects like Friend.tech stems from a misunderstanding of social media's core nature. It applies Marshall McLuhan's theory of "hot" and "cool" media. "Cool" media (like traditional social networks) rely on low-resolution, incomplete signals (e.g., a tweet) that require user participation to create meaning. "Hot" media (like radio or print) deliver complete, high-resolution information that encourages passive consumption. SocialFi attempted to layer finance onto social media by making actions like follows and posts directly tradable with visible, real-time prices. However, this financial signal is a definitive "hot" signal. By superimposing it onto the inherently "cool" medium of social interaction, it fundamentally transformed the medium. Users stopped participating socially and instead began allocating capital rationally based on prices. The financial layer consumed the social one, leaving no genuine social substrate when speculation faded. The article extends this analysis to broader platform decay (e.g., Twitter's shift from cool participation to hot performance metrics) and NFTs. NFT platforms, by optimizing collections with real-time floor prices and rarity scores, rapidly "heated up" the traditionally "cool," participation-rich medium of collecting, destroying its cultural essence and leaving only speculative trading. The solution proposed is not to abandon capital in social contexts, but to design for "condensation points"—localized, infrequent financial interfaces (like Substack subscriptions or Patreon memberships) that allow capital to gather without saturating and overheating the core cool medium. The key lesson is that "liquidity is heat"; adding it to a cool medium doesn't enhance it but alters it, often destroying what made it valuable. Successful platforms will be those that introduce capital while meticulously preserving the cool, participatory nature of their underlying medium.

链捕手05/14 09:22

The Construction of SocialFi Originates from a Misreading of Its Own Medium

链捕手05/14 09:22

The Real AI Bubble, You Can't Buy It

The article argues that the real "bubble" in the current AI boom is largely invisible and inaccessible to the average investor. Unlike the 2000 dot-com bubble, where overvalued companies were publicly traded, the most significant value surges and financial risks are occurring in private markets. Core AI companies like OpenAI, Anthropic, xAI, and Databricks have seen valuations skyrocket (e.g., OpenAI's from $157B to $852B in 18 months), but these transactions happen through private secondary sales, not public stock exchanges. These opaque markets create an "anxiety exposure," leading public investors to chase indirect proxies like memory chip or utility stocks. The author highlights how AI wealth extraction has been radically front-loaded. Employees and founders can cash out years before a potential IPO through structured secondary sales, "founder-led secondary" deals, and collateralized loans against private equity. Major tech firms also use "acqui-hires" or technology licensing deals (like Google/Character.AI, Microsoft/Inflection AI) to secure talent and tech without full acquisitions, allowing early exits outside of regulatory scrutiny. Furthermore, the AI infrastructure build-out is compared to the 2008 real estate bubble. Massive data center projects are financed through complex, off-balance-sheet structures involving private credit, joint ventures, and asset-backed securities using GPUs as collateral (e.g., CoreWeave's deals). This creates a "shadow borrowing" system where the stability of future AI demand underpins trillions in debt, posing systemic risks if expectations falter. The recent collapse of SaaS company Pluralsight, financed by major private credit firms, is cited as a warning. The conclusion is that the most dangerous part of the AI bubble isn't in plain sight on public markets; by the time the average investor sees it, the critical wealth transfers have already occurred in private, unregulated spaces.

marsbit05/14 07:10

The Real AI Bubble, You Can't Buy It

marsbit05/14 07:10

One Article to Understand the Profit Pools and Industry Landscape of the AI Storage Hierarchy

**Deciphering the Profit Pools and Industry Landscape of the AI Storage Hierarchy** AI storage architecture can be divided into six distinct layers based on proximity to computing units: 1) On-chip SRAM, 2) HBM, 3) Motherboard DRAM, 4) CXL pooling layer, 5) Enterprise SSD, and 6) NAS & Cloud Object Storage. In 2025, the total market for these layers (excluding embedded SRAM value) was approximately $229 billion, with DRAM constituting half, HBM 15%, and SSD 11%. The profit landscape is highly concentrated, with over 90% market share in the top three layers for key players. These profit pools are categorized into three types: 1) High-margin, oligopolistic silicon layers (HBM, embedded SRAM, QLC SSD), 2) High-margin, emerging interconnect layers (CXL), and 3) Scalable, recurring-revenue service layers (NAS, Cloud Object Storage). **Key Layers Analysis:** * **On-chip SRAM:** Profits accrue primarily to TSMC via advanced wafer sales for AI chips. * **HBM:** The largest AI-era profit pool, driven by AI accelerator demand. SK Hynix (57-62% share), Samsung, and Micron dominate. HBM boasts exceptionally high margins (e.g., SK Hynix's 72% operating margin in Q1 2026) and is projected to grow at a ~40% CAGR to $100 billion by 2028. * **Motherboard DRAM:** The largest market by revenue ($121.8B in 2025), controlled by Samsung, SK Hynix, and Micron. High profitability is sustained as capacity shifts to HBM. * **CXL Pooling Layer:** Enables rack-level memory sharing for AI workloads. The market is forecast to grow from $1.6B in 2024 to $23.7B by 2033. While memory giants lead, companies like Astera Labs (holding ~55% share in retimers/controllers) achieve very high margins (~76%). * **Enterprise SSD:** A major beneficiary of the AI inference era, especially QLC SSDs, with the market expected to reach $76B by 2030. Samsung, SK Hynix (including Solidigm), and Micron are key players. * **NAS & Cloud Object Storage:** The outermost data lake layer, growing steadily (CAGR ~16-17%). Profit derives from long-term data hosting, egress fees, and ecosystem lock-in, led by vendors like NetApp, Dell, and cloud providers (AWS, Azure, Google Cloud). **Summary:** Profitability correlates strongly with proximity to compute: layers like HBM and CXL components command the highest margins (60%+ and 76%+, respectively) despite smaller market sizes, while DRAM has the largest revenue base. The primary growth vectors are HBM (CAGR ~28%), Enterprise SSD (CAGR ~24%), and CXL pooling (CAGR ~37%). Barriers vary by layer, encompassing advanced manufacturing (HBM), IP/certification (CXL), and high switching costs (service layers).

marsbit05/14 04:03

One Article to Understand the Profit Pools and Industry Landscape of the AI Storage Hierarchy

marsbit05/14 04:03

AI Agents Can Be Verified, But Who Protects Their Privacy?

As AI Agents evolve from automated tools into active participants in on-chain economies, a critical challenge emerges: establishing trust while preserving privacy. While standards like ERC-8004 aim to provide verifiable identity and reputation for agents, their public nature could expose sensitive operational strategies, user preferences, and business relationships in fields like DeFi, governance, and prediction markets. The proposed ACTA (Anonymous Credentials for Trustless Agents) framework addresses this by adding a privacy layer. It allows agents to cryptographically prove they meet certain criteria (e.g., having passed an audit or possessing sufficient reputation) without revealing the underlying sensitive data, using zero-knowledge proofs. This shifts trust from "public identity" to "policy-based proof." This shift is crucial because agents act dynamically on behalf of users, making their behavior a potential proxy for user intent. ACTA would enable verification of an agent's legitimacy or authorization without creating a permanent, public map of all its activities and relationships. ACTA remains a research direction with open challenges, including scalability, decentralization of credential issuers, and implementation costs. However, it highlights a fundamental need: a robust Agent economy requires not just mechanisms for verification, but also for protecting the privacy of agents, their users, and the protocols they interact with.

marsbit05/14 01:27

AI Agents Can Be Verified, But Who Protects Their Privacy?

marsbit05/14 01:27

Circle's Three-Dimensional Valuation Framework: Where Is the Bottom, Where Is the Top

"Circle's 3D Valuation Framework: Where is the Bottom, Where is the Top?" - Article Summary The article analyzes Circle's valuation following its Q1 2026 earnings. While its core business generates substantial interest income from USDC reserves ($6.53B in Q1, up 17% YoY), this revenue is highly sensitive to interest rates and shared significantly with Coinbase. The author proposes a three-dimensional valuation framework: 1. **Interest Business (The Floor):** Valued like a bank (8-15x P/E) on net interest income after Coinbase's share. This provides a conservative valuation baseline. 2. **Payment & Platform Business (The Inflection Point):** Includes CPN (Circle Payments Network) and "Other Revenue" (transaction, integration services). This high-growth segment, not shared with Coinbase, is valued on a platform/network model (higher P/S multiples), similar to Visa/Mastercard. It represents Circle's shift beyond pure interest income. 3. **Arc Network & ARC Token (The Future / Optionality):** Arc is an institutional-focused, EVM-compatible L1 blockchain where USDC is the native gas token. A $222M ARC token pre-sale at a $3B FDV attracted major traditional finance players (BlackRock, Apollo, ICE). While Circle holds 25% of ARC tokens, their value is separate from CRCL equity. This dimension represents the long-term, high-upside bet on Circle becoming an "economic operating system." Current market cap (~$30B) prices in significant future growth beyond the sum-of-the-parts valuation derived from current earnings. The investment thesis hinges on believing in Circle's transition from a "stablecoin issuer" to a broader financial infrastructure and network platform. Key variables for the future include USDC adoption growth, CPN network effects, Arc's success, and potential renegotiation of the Coinbase revenue-sharing agreement.

marsbit05/13 13:56

Circle's Three-Dimensional Valuation Framework: Where Is the Bottom, Where Is the Top

marsbit05/13 13:56

From Gas Limit to 'Keyed Nonces', How to Understand the Next Step in Ethereum Scalability?

Ethereum’s scalability efforts are shifting toward a user-centric approach—focusing not only on higher TPS, but on translating technical upgrades into lower costs, smoother operations, and better wallet experiences. Two recent developments highlight this direction: - **Raising the Gas Limit to 200 million**: Following the Fusaka upgrade that increased it to 60 million, a consensus has formed around a potential future increase to 200 million. This would boost Ethereum’s execution capacity, but it is planned alongside other upgrades—such as ePBS, Block-Level Access Lists (BAL), and EIP-8037—to manage state growth and keep node operation viable for average participants. - **Keyed Nonces (EIP-8250)**: This proposal aims to improve how transactions are queued. Instead of a single linear nonce per account, it introduces multiple independent nonce domains. This prevents different types of transactions—such as private payments, session keys, or batch operations—from blocking each other. Vitalik Buterin views this as a foundational step toward better privacy support and more flexible state scalability. Together, these upgrades are part of a broader move to push complexity from wallets, DApps, and relays back into the protocol layer. For everyday users, this means future Ethereum interactions could become less congested, more intuitive, and safer—especially as core improvements in account abstraction, cross-L2 interoperability, and node decentralization continue to progress. Ultimately, Ethereum is evolving to handle not just more transactions, but more varied and complex on-chain use cases while preserving its decentralized foundation.

marsbit05/13 09:17

From Gas Limit to 'Keyed Nonces', How to Understand the Next Step in Ethereum Scalability?

marsbit05/13 09:17

活动图片