2026-05-23 Суббота

Новостной центр - Страница 24

Получайте криптоновости и тенденции рынка в режиме реального времени с помощью Новостного центра HTX.

Bernstein's 97-Page Report Decoded: The Battle for AI Data Center Connectivity, Who Will Be the True Winner by 2026?

Bernstein's 97-page report analyzes the AI data center connectivity landscape. It argues that the bottleneck is shifting from raw compute (GPU) to the systems connecting GPUs, crucial for cluster efficiency. Copper and optical interconnects are not in a simple replacement cycle but will coexist long-term, with copper dominating short-distance "scale-up" connections and optics favored for longer "scale-out" scenarios. While Co-Packaged Optics (CPO) is the long-term direction for power and cost savings, its widespread adoption faces manufacturing and reliability hurdles, with mass deployment unlikely before 2028. Transitional technologies like Linear Pluggable Optics (LPO) and Near-Packaged Optics (NPO) are seen as near-term leaders. A key insight is that CPO will fundamentally reshape the value chain, shifting profits from traditional optical module suppliers towards chip designers (e.g., NVIDIA, Broadcom), advanced packaging (e.g., TSMC), and system integrators. For 2026, the report highlights more immediate and certain investment opportunities in the essential "infrastructure" enabling this connectivity shift. This includes upgrades for PCBs, ABF substrates, and CCLa driven by new AI server/switch platforms, alongside demand for 1.6T optical modules, LPO/NPO, and the testing/validation equipment required for future CPO scale-up.

marsbit05/19 03:16

Bernstein's 97-Page Report Decoded: The Battle for AI Data Center Connectivity, Who Will Be the True Winner by 2026?

marsbit05/19 03:16

Understanding the New Economic Model of Tokenization

Understanding the New Token Economics Model The commercialization of AI applications is evolving from selling software and subscriptions to selling token call capacity. Tokens, the fundamental unit of information processing for large language models (LLMs), have become the basis for API billing and consumption. With call volumes exploding, tokens themselves are now being traded—procured, routed, split, and resold—forming a new intermediary market. This layer connects upstream LLM providers with downstream developers and enterprises, acting as a global wholesale-to-retail liquidity network. The rise of this business is fueled by a massive surge in China's daily token call volume—growing over a thousandfold from 100 billion in early 2024 to over 140 trillion by March 2026—and significant improvements in domestic LLM capabilities, which are now competitive globally. The core value of token distribution platforms extends beyond simple arbitrage. Key functions include aggregating multiple models (like GPT, Claude, and domestic models such as Kimi and DeepSeek) under a unified API, lowering network and payment barriers, and providing enterprise services like model selection, prompt engineering, and system integration. Profit models are diversifying: (1) resale margins; (2) technical premiums from proprietary inference acceleration (e.g., reducing costs to 1/10 of the industry standard); and (3) enterprise value-added services. High-consumption scenarios like marketing, short-form video, gaming, and e-commerce are primary drivers. Investment opportunities are seen in both companies with strong model capabilities (e.g., Alibaba, Tencent, MiniMax) and those with high-consumption client scenarios (e.g., marketing agencies with overseas reach). However, risks are significant: low entry barriers leading to intense competition, capital requirements and bad debt risks from advance payments, and dependency on policy changes from upstream LLM providers who control API pricing and access.

marsbit05/19 02:54

Understanding the New Economic Model of Tokenization

marsbit05/19 02:54

Farewell to the Copper Era: Understanding the Logic of the AI Silicon Photonics Industry Chain and Key US Stock Players

**Summary: The Era of Silicon Photonics and Key AI Infrastructure Stocks** The article delves into the transition from copper-based interconnects to silicon photonics (SiPh) as a critical enabler for next-generation AI data centers. It explains that copper faces fundamental physical limits—the bandwidth wall, density wall, and power wall—at high data rates (1.6T+), making a material shift essential. Silicon photonics, which integrates components like lasers, modulators, and detectors onto a silicon chip, offers a solution by leveraging mature CMOS manufacturing for cost-effective, high-volume production. A key challenge is that silicon itself is not an efficient light source, making Indium Phosphide (InP) lasers a critical and supply-constrained component. A major industry catalyst was NVIDIA's 2025 GTC announcement, declaring optical interconnects a "standard" from its Rubin platform onward, followed by strategic investments to secure the supply chain. The industry is structured in four key layers: 1. **Foundries:** TSMC leads with its COUPE platform, while Tower Semiconductor (specialized SiPh foundry) and GlobalFoundries are major players. 2. **Core Component Suppliers:** Lumentum is highlighted as the sole volume manufacturer of the crucial 200G/lane EML laser, with orders locked by NVIDIA through 2027. 3. **Module & System Manufacturers:** Coherent holds significant market share, with Chinese manufacturers like InnoLight also noted for scale. 4. **System Integrators:** NVIDIA, Broadcom, and Marvell dominate this layer, setting standards and integrating technology. The article identifies core public investment targets: **NVIDIA (NVDA)** as the ecosystem driver; **Broadcom (AVGO)** and **Marvell (MRVL)** in networking/switching chips; **Lumentum (LITE)** and **Coherent (COHR)** for critical components; and foundries **TSMC (TSM)** and **Tower Semiconductor (TSEM)**. Private companies Lightmatter and Ayar Labs are noted as key IPO candidates. The silicon photonics shift is driving a re-rating of company valuations, moving them from traditional telecom/industrial metrics to premium AI infrastructure multiples. The industry features high barriers to entry (e.g., multi-year lead times for InP laser capacity, complex 3D integration/thermal management, and lengthy customer qualification cycles), suggesting a "winner-takes-most" dynamic. Risks include dependence on hyperscaler capex cycles, potential technology disruption among competing optical approaches (LPO, CPO, OCS, Optical I/O), and a timeline where widespread CPO deployment may not occur until ~2028, with LPO serving as a transitional technology. The conclusion advises that betting on the overall industry trend may be safer than betting on any single company.

marsbit05/19 02:15

Farewell to the Copper Era: Understanding the Logic of the AI Silicon Photonics Industry Chain and Key US Stock Players

marsbit05/19 02:15

POLY's Appearance Hints Are Getting Denser, How Far Away Is the Polymarket Airdrop?

**POLY Debut Hints Grow More Frequent: How Far is the Polymarket Airdrop?** Recent continuous hints from Polymarket team members regarding the POLY token have sparked widespread analysis within airdrop communities about its launch timeline and potential scale. According to predict.fun data, the probability of "Polymarket launching its official token before year-end 2025" currently stands at 56%. Hints about POLY began in October 2024. CEO Shayne Coplan's social media post mentioning $POLY alongside major cryptocurrencies first fueled speculation. This was followed by Growth Lead William LeGate discussing "prospective airdrop farmers," and CMO Matthew Modabber explicitly confirming that "Polymarket will have a token, and there will be an airdrop." The momentum continued into 2025. In April, major crypto data platforms CoinGecko and CoinMarketCap created placeholder pages for POLY, further solidifying expectations. May saw discussions shift towards token utility and airdrop criteria. A team member's "Soon" reply to a question about staking POLY for fee reductions, and a leaked internal screenshot showing an "Airdrop" tab, significantly increased anticipation. LeGate also outlined potential airdrop qualifiers, mentioning badges for employees, high-volume/high-profit traders, and ecosystem builders. He suggested that linking a Polymarket account to X (Twitter), sharing trades and market insights, and actively engaging with the community might constitute part of the eligibility criteria, leading to a surge in related social media activity. Despite the growing hype, the article's author expresses a personal view that Polymarket might prioritize platform stability and infrastructure upgrades—especially with the upcoming World Cup—over an immediate token launch, opting for a post-event release. Their current strategy involves hedging positions across prediction markets, actively trading on Polymarket, and engaging on social media to potentially qualify for a future airdrop.

Odaily星球日报05/19 01:46

POLY's Appearance Hints Are Getting Denser, How Far Away Is the Polymarket Airdrop?

Odaily星球日报05/19 01:46

活动图片