# 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.

Investment Philosophy of Gavin Baker, an Early Nvidia Investor: Long AI Infrastructure Bottlenecks, Short Overall Market Risk

Gavin Baker, an early investor in Nvidia and founder of Atreides Management, outlines his investment philosophy: going long on AI infrastructure bottlenecks while hedging against broader market risk. He argues AI is not a bubble but a supercycle driven by constraints in power, wafers (semiconductors), and compute efficiency (tokens per watt). True alpha, he believes, lies not in application-layer companies like OpenAI but in "picks and shovels" providers—companies solving physical bottlenecks in GPU connectivity (e.g., Astera Labs), memory (Micron), inference chips (Cerebras, Positron), advanced manufacturing (TSMC, ASML), and energy supply. His portfolio reflects this barbell strategy: concentrated bets on key infrastructure players alongside a significant put position on the QQQ ETF to hedge overall market downside. Baker contends this cycle differs from the dot-com bubble because demand is fueled by the strong balance sheets of hyperscalers (Google, Meta, Amazon, Microsoft), not debt, and physical supply constraints (e.g., chip manufacturing capacity) prevent runaway overinvestment. He highlights the growing importance of inference (vs. pre-training), vertical/small language models, sovereign infrastructure deployment speed, and the convergence of energy and space (e.g., orbital compute). His long-term view is that performance-per-watt and token cost reduction will dictate winners as AI scaling hits fundamental physical limits.

marsbit05/30 03:23

Investment Philosophy of Gavin Baker, an Early Nvidia Investor: Long AI Infrastructure Bottlenecks, Short Overall Market Risk

marsbit05/30 03:23

Apple Re-invented Image Compression with AI: Same Quality, One-Third the File Size

Apple’s PICO: An AI-Powered Image Codec That Cuts File Size by Two-Thirds at Equal Perceived Quality In 2025, JPEG AI became the first international standard for learned image compression. However, it, like most codecs, still prioritizes mathematical metrics like PSNR over true perceptual quality—what the human eye finds pleasing. Apple researchers have introduced PICO (Perceptual Image Codec), a neural codec designed to optimize for human perception. It tackles key practical challenges: 1) Speed: A novel "one-shot context model" accelerates entropy encoding without sacrificing compression efficiency. 2) Artifacts: A dedicated TextFidelity loss preserves text clarity, and a TilingArtifact loss eliminates color seams between image tiles processed in parallel. 3) Control: It avoids the "hallucinations" common in GAN-based perceptual models. In a large-scale human evaluation (74,925 comparisons), PICO achieved the same perceived quality as standards like AV1, VVC, and JPEG AI while using only 30-43% of the bitrate. It also outperforms other learned perceptual codecs by 20-40%. Remarkably, it runs in 230ms (encode) and 150ms (decode) on an iPhone 17 Pro Max. While less efficient on synthetic graphics, PICO represents a significant shift from optimizing mathematical scores to directly targeting human visual experience, making high-quality perceptual compression practical for consumer devices. The work builds on expertise from WaveOne, whose team joined Apple and previously advanced neural video compression.

marsbit05/30 02:47

Apple Re-invented Image Compression with AI: Same Quality, One-Third the File Size

marsbit05/30 02:47

Shanghai's Leading Large Model Company Initiates A-Share Listing

Shanghai-based AI large language model leader MiniMax has initiated the process for an A-share listing in China, having filed a pre-IPO tutoring report with the Shanghai Securities Regulatory Bureau on May 29. This move positions it to compete with Zhipu AI for the title of the first major domestic LLM company to list on the A-share market. Having already completed an IPO in Hong Kong in January 2026, MiniMax's stock price has surged approximately 409% since its debut, with its market capitalization reaching around HK$263.45 billion (approximately RMB 227.55 billion) as of May 29. The company's rapid growth is supported by strong business performance. Its Annual Recurring Revenue (ARR) has grown over 100% in the past two months and now exceeds $300 million. It serves over one million global enterprise and developer clients and has around 300 million users worldwide. For the full year 2025, MiniMax reported revenue of $79.038 million, with a gross margin of 25.4%. While it reported an adjusted net loss of $250 million, the loss rate has narrowed significantly year-over-year. On the product front, MiniMax has released several flagship models this year, including MiniMax-M2.5, M2.6, and M2.7, with the first and last being open-sourced. Its models gained significant traction earlier in the year, briefly becoming the top model provider by usage share on the OpenRouter platform in February. The company has also upgraded its AI agent product, now named Mavis, and is preparing to launch its next-generation MiniMax-M3 model. Technical previews indicate M3 will feature a novel "MiniMax Sparse Attention" mechanism, promising substantial improvements in inference speed. MiniMax's push for an A-share listing reflects a broader trend among China's leading AI firms, including Zhipu AI, Moonshot AI, StepFun, and 01.AI, to seek public listings. This strategy aims to secure broader financing channels to support the immense computational costs and ongoing commercialization efforts inherent in developing advanced large language models.

marsbit05/30 02:45

Shanghai's Leading Large Model Company Initiates A-Share Listing

marsbit05/30 02:45

Claude Opus 4.8 Released, Anthropic Begins to Market "Trustworthiness" as a Product Feature

Anthropic has released Claude Opus 4.8, emphasizing "trustworthiness" as a core selling point alongside performance gains. While it leads in five of six key benchmarks, the most significant improvement is a dramatic reduction in the model's failure to report its own errors in code tasks, cutting the rate from 19.7% to 3.7%. This increased reliability and self-awareness is positioned as crucial for deploying AI in real-world workflows. Key updates include enhanced mathematical and long-context reasoning, the introduction of dynamic multi-agent workflows in Claude Code for automated verification, and improved token efficiency. However, the model shows some regressions in areas like resisting prompt injection. Pricing remains unchanged. The release also signals the impending arrival of Anthropic's more powerful, restricted "Mythos"-tier model in the coming weeks. The article frames the industry shift as moving beyond pure benchmark competition towards a focus on reliability and verifiability, enabling users to delegate more critical tasks to AI agents with greater confidence.

marsbit05/29 22:19

Claude Opus 4.8 Released, Anthropic Begins to Market "Trustworthiness" as a Product Feature

marsbit05/29 22:19

Snowflake's stock price surges 33%, AI infrastructure expands from chips to the data layer

Snowflake's stock surged over 33% after the company raised its full-year revenue forecast and announced a $6 billion, five-year partnership with AWS. This agreement, which secures AWS Graviton chip supply, underscores Snowflake's evolving role in the AI infrastructure landscape. The market's reaction reflects a reassessment of data platforms as critical enablers for enterprise AI adoption. As companies move beyond conceptual AI to building operational, data-integrated AI workflows, demand for robust data storage, processing, and analysis capabilities is increasing. The deal strengthens AWS's custom chip ecosystem while positioning Snowflake not just as a data warehouse, but as a key data layer for deploying and managing AI applications. Analysts have rapidly revised their valuations upwards, signaling a shift in sentiment where proven AI-driven revenue growth is now commanding premium multiples.

marsbit05/29 19:09

Snowflake's stock price surges 33%, AI infrastructure expands from chips to the data layer

marsbit05/29 19:09

Biology's Paradigm Shift: Zuckerberg's New Open-Source Model Completely Overturns Google's AlphaFold Throne

The AlphaFold era faces a major challenge. A new open-source AI model, ESMFold2, from Meta CEO Mark Zuckerberg's Biohub, has been released alongside a massive database of 11 billion predicted protein structures—surpassing the AlphaFold database by 8 billion entries. Published in Nature, the model is reported to outperform AlphaFold3 in key areas, particularly in predicting protein complexes. Crucially, it is fully open-source with no commercial restrictions. ESMFold2 takes a different technical approach, building on a protein language model trained on billions of sequences, including microbial data from diverse environments like soil and ocean—areas less covered by AlphaFold. The team validated its utility by designing and successfully synthesizing novel functional proteins in the lab. The decision to open-source everything is seen as a strategic move, similar to Meta's approach with its Llama models, aiming to build an ecosystem and accelerate global research. While scientists welcome the resource, some urge caution, noting the need for independent validation of predictions and questioning its performance on entirely novel protein folds. The development signals intensified competition in protein AI, rapidly evolving much like the large language model field, and represents a significant step forward in using AI to decode and engineer the machinery of life.

marsbit05/29 12:31

Biology's Paradigm Shift: Zuckerberg's New Open-Source Model Completely Overturns Google's AlphaFold Throne

marsbit05/29 12:31

VVV Surges Over 10x This Year, Is the Base Ecosystem the Last Hope for Crypto AI?

The article highlights the surge of AI concepts within the Base blockchain ecosystem, positioning it as a focal point for crypto-AI activity. It spotlights several key projects driving this trend. Venice AI's token VVV is noted as a leader, having surged over 10x year-to-date, supported by its dual-token model with DIEM and real business revenue for buybacks. Other significant projects include Virtual, a major AI agent launchpad; Clanker, an AI token launch platform originating from Farcaster; Freysa AI, experimenting with sovereign AI agents; Elsa, an AI-powered DeFi execution layer; and Warden Protocol, a modular L1 for an "agent internet." The piece observes that Base has become a rare active hub for crypto-AI, fueled by the growth of AI agents and related payment ecosystems. While mentioning the volatility of AI-themed meme coins on Base, the article concludes that the ecosystem's potential to realize a fully "on-chain AI world" remains an open question.

marsbit05/29 11:09

VVV Surges Over 10x This Year, Is the Base Ecosystem the Last Hope for Crypto AI?

marsbit05/29 11:09

VVV Skyrockets Over 1000% Year-to-Date, Is Base Ecosystem the Last Hope for Crypto AI?

VVV Surges Over 10x This Year: Is Base Ecosystem the Final Hope for Crypto AI? The AI wave continues, and within the crypto space, the Base ecosystem is emerging as a key hub for AI concepts. Beyond VVV's impressive 1076% yearly gain, other projects like Virtual and Clanker are making steady progress. Infrastructure for AI Agent payments, such as the x402 protocol, is developing, and platforms related to L1s, operating systems, wallets, and social networks for AI Agents are also appearing. Key projects highlighted include: - **VVV (Venice)**: The leading AI token on Base, it operates a dual-token model with compute token DIEM. Its price, supported by real revenue from the privacy-focused Venice AI platform, recently hit around $18 before settling near $16. - **VIRTUAL**: A top Base launchpad positioning itself as an AI Agent co-ownership layer. It supports token creation and monetization for autonomous AI Agents. - **Clanker**: An AI launchpad originating from Farcaster that allows token creation via social media posts. - **FAI (Freysa AI)**: An experiment in creating a "Sovereign AI Agent" that autonomously controls its crypto assets. - **ELSA**: An AI execution layer for DeFi, translating natural language into on-chain actions. - **WARD (Warden Protocol)**: A modular L1/OS for a decentralized "internet of agents." The summary also mentions the volatility of AI-themed meme coins on Base. While Base has become a notably active ecosystem for crypto AI, driven by AI Agent development and payment solutions, it remains uncertain whether it can fully realize the vision of an "on-chain AI world."

Odaily星球日报05/29 11:09

VVV Skyrockets Over 1000% Year-to-Date, Is Base Ecosystem the Last Hope for Crypto AI?

Odaily星球日报05/29 11:09

AI Is Not Replicating the Internet; It’s Replicating the Industrial Revolution

AI is not replicating the Internet; it is replicating the Industrial Revolution. The past two decades of the internet were built on monetizing user attention and ad space. In contrast, the current AI commercialization path reveals a clear structural shift: the focus is moving from serving consumers (C端) to replacing human labor costs for businesses (B端). While C端 AI applications like ChatGPT face stagnant subscription growth and low conversion rates (often below 5%), the B端 market is exploding. Anthropic's annualized revenue soared from $90 billion to $450 billion in early 2026, primarily driven by enterprise API and Agent deployments. The core logic is Return on Investment (ROI): companies spend on AI to save significantly more on salary costs. For instance, an AI coding agent can replace hundreds of junior programmers, offering a clear and compelling cost-benefit equation. The fundamental mismatch lies in the underlying business logic. C端 AI struggles due to low user switching costs, lack of network effects, and an inability to capture significant user time like entertainment apps. Conversely, B端 AI thrives because enterprises buy based on measurable ROI, integrate AI deeply into workflows (creating high switching costs), and are willing to pay a premium for stability and performance. AI is evolving from a digital tool into a digital labor force—directly executing tasks rather than just assisting humans. This transformation mirrors the Industrial Revolution, where machinery replaced physical labor. Today, AI is replacing structured cognitive labor. The total global wage bill represents a market vastly larger than internet advertising. Therefore, the true value of AI lies not in capturing traffic, but in capturing the economics of labor cost replacement. The internet monetized attention; AI monetizes wages.

marsbit05/29 10:24

AI Is Not Replicating the Internet; It’s Replicating the Industrial Revolution

marsbit05/29 10:24

Deconstructing the Investment Philosophy of Gavin Baker, Early Nvidia Investor: Long AI Infrastructure Bottlenecks, Short Overall Market Risk

**Summary: Gavin Baker's AI Infrastructure Investment Philosophy** Gavin Baker, an early investor in Nvidia and Cerebras, views AI not as a bubble but as a "supercycle" driven by fundamental infrastructure bottlenecks. He argues the greatest returns lie not in AI applications (like chatbots) but in the "picks and shovels" layer: the physical constraints of power, semiconductor wafers, and compute. His core investment thesis focuses on "performance per watt" — companies that lower the cost per AI-generated token. Key bottlenecks he targets include GPU interconnects (e.g., Astera Labs), memory (Micron), inference chips (Cerebras, Positron), advanced chip manufacturing (TSMC, ASML), and power supply. He also sees value in 3D world-model builders (Unity) for future AI/robotics training and in sovereign infrastructure deployment speed. While bullish on specific AI infrastructure winners, Baker hedges overall market risk by holding QQQ put options. He differentiates the current cycle from the dot-com bubble, arguing it's funded by the strong cash reserves of tech giants, not debt, and is physically constrained by slow-to-scale supply chains (like chip fab capacity and electricity grids), which prevent a runaway bubble.

marsbit05/29 08:33

Deconstructing the Investment Philosophy of Gavin Baker, Early Nvidia Investor: Long AI Infrastructure Bottlenecks, Short Overall Market Risk