# Сопутствующие статьи по теме AI

Новостной центр HTX предлагает последние статьи и углубленный анализ по "AI", охватывающие рыночные тренды, новости проектов, развитие технологий и политику регулирования в криптоиндустрии.

SpaceX, OpenAI, Anthropic: The Three AI Giants Racing for IPO, Which One Is Worth Betting On?

SpaceX, OpenAI, and Anthropic are poised for historic IPOs within weeks, potentially raising a combined $180 billion—a sum exceeding the entire internet bubble's fundraising. The hosts of the Limitless Podcast argue this isn't just individual company financing but an unprecedented capital concentration for AI infrastructure, driven by an insatiable need for compute, data centers, power, and chips. SpaceX's IPO is notable for reportedly changing market index rules to allow faster inclusion, potentially funneling trillions in passive retirement funds into its stock, despite its unproven space-based data center business model. In contrast, Anthropic demonstrates explosive growth, with ARR reportedly hitting $45 billion and approaching profitability, fueled by strong enterprise adoption of products like Claude Code. Google's separate $80 billion raise highlights the immense capital pressure, even for giants. The discussion acknowledges bubble risks but leans optimistic. The hosts contend the massive spending is building essential physical infrastructure for the next technological era. A key bottleneck isn't capital but the real-world limits of chip manufacturing and construction speed. As long as demand for AI compute outstrips supply, this investment cycle represents a foundational build-out rather than a purely financial bubble. All three companies are seen as foundational bets on the future, with Anthropic often cited as the most immediately compelling due to its proven revenue trajectory.

marsbit06/04 11:47

SpaceX, OpenAI, Anthropic: The Three AI Giants Racing for IPO, Which One Is Worth Betting On?

marsbit06/04 11:47

The Merger of Codex and ChatGPT Marks the Beginning of a Major Reshuffle in Programming Tools

OpenAI is shifting its strategic focus from ChatGPT to Codex, merging them along with the browser tool Atlas into a unified desktop super-app. This move signals an internal belief that Codex, originally a programming tool, represents the next evolution of AI more than conversational models like ChatGPT. Over the past year, Codex's weekly active users have surged past 5 million. The key distinction is that while ChatGPT answers questions, Codex executes tasks. Enterprises increasingly value this ability to get work done over simply receiving advice. Consequently, Codex is attracting professionals beyond developers, including analysts, bankers, marketers, and product managers. OpenAI's reorganization and increased investment in Codex stem from recognizing that the future of AI competition lies in execution capabilities, not just conversation. The company is launching role-specific plugins (e.g., for data analysis, sales, design) to transform Codex into a broad knowledge work platform that automates and redefines white-collar workflows. Beyond being a tool, Codex reflects OpenAI's ambition to redefine software. New features like "Sites"—which generates interactive websites from documents—and collaborative "Annotations" aim to create a paradigm where the AI understands the goal and handles the tools and steps, functioning more like a digital colleague than traditional software. The ultimate goal is a unified experience where the user cares only about the completed task.

marsbit06/04 11:32

The Merger of Codex and ChatGPT Marks the Beginning of a Major Reshuffle in Programming Tools

marsbit06/04 11:32

GitHub, Transfixed by AI

On the night of February 9th, GitHub suffered a major outage caused by a simple configuration change—reducing a cache refresh interval from 12 to 2 hours—that triggered a cascade of failures. This was not an isolated event, but part of a broader pattern. In early 2026, GitHub experienced at least 8 major incidents, failing to meet its promised 99.9% availability. These outages stemmed from structural issues: explosive growth in load, tight service coupling, and insufficient protection against abnormal traffic. This unprecedented load is driven by AI Agents. In 2025, GitHub handled ~1 billion commits. By 2026, weekly commits reached 275 million, projecting to ~14 billion for the year—a 14x increase. AI tools like Claude Code now contribute 4.5% of all public repository commits, with weekly submissions surging 25x in just three months. AI-generated pull requests jumped from 4 million to 17 million per month in half a year. Unlike human developers, AI Agents work continuously, generating commits at a scale that overwhelms infrastructure designed for human rhythms. The surge also shattered GitHub's business model. Copilot's flat-rate pricing, based on assisting human developers, became unsustainable as Agentic AI sessions consumed resources worth hundreds of dollars for a few dollars in fees. In response, GitHub imposed usage limits and, by June 1st, shifted to a pay-per-use "AI Credits" system. Facing this new reality, GitHub realized a 10x scaling plan was insufficient. It announced a need to *redesign* its architecture for 30x current scale—decoupling services, adding fault isolation, and improving change management to prevent cascading failures. Other platforms like Stripe and AWS are facing similar challenges with AI Agents. Fundamentally, GitHub is transitioning from a human collaboration platform to an "exhaust pipe" for automated AI workflows. Its detailed post-mortem reports aim to maintain trust during this turbulent rebuild. The February outage was not just a technical glitch, but a signal of the software industry's entry into a new, AI-driven era.

marsbit06/04 10:40

GitHub, Transfixed by AI

marsbit06/04 10:40

Where the AI Bubble Really Is: Which Layer of Players Are Naked

AI Bubble: Where It Really Is and Who's Swimming Naked This analysis dissects the AI industry not as a single entity but as a five-layer pyramid, arguing that bubbles are concentrated in specific tiers, not uniformly distributed. **Key Distinction from the 2000 Dot-com Bubble:** Unlike 2000, where companies had stock prices before revenue, today's leading AI players have massive, contract-backed revenue driving their valuations. Core infrastructure demand is real, with every GPU running at full capacity for paying customers. **The Five-Layer Pyramid & Bubble Assessment:** * **L0 (Fab/Manufacturing) & Top L4 (Leading AI Apps): NO BUBBLE.** Companies like TSMC, NVIDIA, major cloud providers (Microsoft, Google, Meta, Amazon), and top AI labs have real revenues and orders. Supply is tightly constrained by TSMC's disciplined capacity control and physical limits like power/land for data centers, preventing a supply glut. * **L1 (Memory): BATTLEGROUND.** Sky-high HBM margins could signal a new structural cycle or a classic "boom before bust." The oligopoly of three major players may enforce supply discipline, making this a high-stakes bet. * **L2 (Interconnect/Optical Modules): BUBBLE TERRITORY.** Companies like Lumentum and AAOI have seen stock surges (4-10x) far outpacing revenue growth. This hardware segment has lower physical barriers to expansion than fabs, allowing speculation. It mirrors the 2000 bubble's epicenter—optics. * **L3 (Infrastructure/"GPU Landlords"): VULNERABLE.** GPU leasing companies profit from the current compute shortage but own no long-term moat. Their business model relies on a temporary bottleneck that will ease as big tech expands and new tech (e.g., potential space-based data centers) emerges. * **L4 Long Tail (VC-backed Startups): STRONG BUBBLE SIGNALS.** VC funding concentration in AI is twice that of the 1999 peak. Many startups with little revenue use the valuation logic of successful giants to justify their own, creating high risk of a "valuation crunch" when funding dries up. **Critical Risks to Monitor:** 1. **GPU Depreciation & Accounting:** Companies extending the assumed useful life of GPUs artificially boost profits. The true economic life depends on future generational leaps from NVIDIA. 2. **"GPU Credit" & Off-Balance-Sheet Leverage:** Emerging structures where shell companies borrow to buy GPUs and lease them out (with chipmakers sometimes investing) move debt off major balance sheets. This echoes the "vendor financing" of 2000 and the securitization risks of 2008, though currently small-scale. 3. **TSMC Abandoning Caution:** If the primary supply bottleneck (TSMC's conservative capacity planning) breaks, runaway supply could trigger a bust. 4. **Algorithmic Efficiency Breakthrough:** A major leap in software efficiency could drastically reduce the need for raw compute hardware, undermining the investment thesis. **Conclusion:** The AI boom is expensive and has frothy areas, but its core is underpinned by real demand and physical supply constraints. The bubble risk is layered: most present in optical components, GPU leasing, and the long-tail startup ecosystem, while the foundational chip manufacturing and leading application layers remain relatively solid—for now.

marsbit06/04 10:20

Where the AI Bubble Really Is: Which Layer of Players Are Naked

marsbit06/04 10:20

Standing in the Light: A Comprehensive Guide to the Optical Module and CPO Supply Chain

"Standing in the Light: Understanding the Optical Module and CPO Industry Chain" This article analyzes the critical role of optical communication technology, specifically optical modules and Co-Packaged Optics (CPO), as the "nervous system" for modern AI data centers. With exponential growth in AI computational demands (e.g., NVIDIA's Vera Rubin architecture), traditional electrical interconnects using copper cables face severe bottlenecks in bandwidth, power consumption, and signal integrity over distance. The core function of an optical module is to act as a "translator," converting electrical signals from chips into optical signals for transmission over fiber (and vice-versa). Key internal components include lasers, modulators, photodetectors, drivers, and DSP chips. The industry is currently transitioning from 800G to 1.6T modules. However, the future lies in CPO. This next-generation technology integrates the optical engine directly with the switch ASIC/XPU on the same package substrate, drastically reducing power consumption (by ~3.5x according to NVIDIA), overcoming bandwidth density limits, and minimizing signal attenuation compared to traditional pluggable modules. Key challenges for CPO include advanced packaging capacity (dominated by TSMC), thermal management, repairability, and standardization. The article details the broader technology landscape, including Near-Packaged Optics (NPO, a pragmatic intermediate step), Linear-drive Pluggable Optics (LPO), Optical I/O (OIO for chip-level integration), and Optical Circuit Switches (OCS). A comprehensive CPO industry chain is mapped, highlighting shifting power dynamics: * **Architecture Definers:** NVIDIA, Broadcom, and Marvell now hold greater influence. * **Advanced Packaging & Manufacturing:** TSMC is central; Fabrinet is a key EMS player. * **Lasers ("The Heart"):** A strategic bottleneck. EML lasers are led by Lumentum and Coherent (both receiving major NVIDIA investments). CW lasers, favored for CPO/silicon photonics, see strong Chinese players like Source Photonics and Sicoya. * **Silicon Photonics Chips:** The mainstream path for CPO engines, with key players like Broadcom, Intel, Marvell, and China's Accelink. * **Fiber Connectivity Components:** A major new, high-growth market created by CPO, including Fiber Array Units (FAU), Polarization-Maintaining Fiber (PMF), and MPO connectors. Companies like Tianfu Communication and US Conec are leaders. * **Fiber & Cable:** Experiencing a super-cycle (e.g., Corning, Yangtze Optical Fiber). * **PCB/Substrates:** Requiring advanced materials (e.g., Shengyi Tech). * **DSP & SerDes:** Functions are integrated into switch ASICs in the CPO era (e.g., Broadcom, Astera Labs). * **Optical Module Makers:** Transitioning from standalone module suppliers to providers of optical engines and NPO/LPO solutions while riding the current pluggable boom (e.g., Zhongji Innolight, Eoptolink). The investment timeline is segmented: Short-term (2026-2027) features the "last feast" for pluggable modules and CPO's initial rollout. Medium-term (2027-2029) will see CPO expand and NPO peak. Long-term (2029-2032+) involves CPO/OIO penetration into intra-rack scaling. In conclusion, optical interconnects are fundamental to AI infrastructure. The competitive landscape sees US firms leading in architecture and high-end chips, TSMC in advanced packaging, and Chinese firms holding strong positions in modules, connectivity components, CW lasers, and fiber/cable. The future belongs to companies that can navigate the technological shift from "selling shovels" (modules) to "building highways" (CPO/OIO infrastructure).

marsbit06/04 10:10

Standing in the Light: A Comprehensive Guide to the Optical Module and CPO Supply Chain

marsbit06/04 10:10

AI Trading Cools, South Korean Stocks Plunge 1.8%, Spot Gold Rises 1%, Bitcoin Dives

A sell-off in AI-related stocks, triggered by Broadcom's disappointing earnings forecast, sent shockwaves through global markets. South Korea's KOSPI led Asia's decline, plunging 1.8% as the risks from concentrated chip stock gains and surging leveraged investments came to the fore. The tech-heavy Nasdaq 100 futures fell 0.5% following Broadcom's 14% after-hours plunge, which signaled a slower-than-expected transition to AI clients. This pullback extended Wall Street's weakness, halting the S&P 500's nine-day rally amid hawkish Fed signals and renewed Middle East tensions. South Korean authorities convened an emergency meeting, pledging "immediate measures" against market volatility and warning of record-high stock margin debt. The adjustment rippled across assets: Bitcoin fell to around $64,000, its lowest since February, while safe-haven gold rose 1% on bargain hunting. Oil prices dipped on Middle East ceasefire news. Market analysts noted the sell-off was driven by profit-taking after massive gains, particularly in chip stocks like Samsung and SK Hynix, which now dominate the KOSPI. Wall Street banks are divided on Korea's outlook, with Goldman Sachs raising its target while Citigroup and others warn of overvaluation and a potential bubble. Bridgewater's Ray Dalio noted that great technological shifts often create bubbles. Meanwhile, Fed officials' hints at potential future rate hikes added to the cautious mood ahead of key U.S. jobs data.

华尔街日报06/04 07:47

AI Trading Cools, South Korean Stocks Plunge 1.8%, Spot Gold Rises 1%, Bitcoin Dives

华尔街日报06/04 07:47

活动图片