Tag
MIT Technology Review examines the overlooked data storage costs of AI inference, noting that a single 8-second AI-generated video can leave behind 500MB of back-end data (130x the user-visible file) and citing IDC/McKinsey forecasts of 35% CAGR inference demand growth through 2030.
An analysis of why Microsoft, despite its vast compute, data, and engineering talent, has yet to successfully develop its own frontier AI model.
一份基于可核查数据的深度分析，统计了自 ChatGPT 发布以来被指“被 AI 杀死”的 56 项业务/行业：6 项消失、22 项萎缩、10 项转型、13 项反而增长，指出许多 AI 冲击的说法经不起数据检验。
一条观点性推文，认为编码AI正快速商品化、Anthropic长期前景不佳，而OpenAI聚焦消费级AI与硬件是更聪明的战略。
We checked the bios and departure reports for the 72 individuals publicly named by Meta's Superintelligence Labs: 40 still list Meta, 9 have left, and 23 couldn't be verified. Over 80% of those verified remain, though two of the three highest-paid hires (Tulloch, Ruoming Pang) have departed.
一则推文线程，通过将硅谷与北京、上海、杭州、深圳的机器人公司绘制在5张地图上，对比各城市的机器人产业规模与结构差异：硅谷67家、深圳36家、北京32家、上海25家、杭州25家。
作者质疑美国实验室为何似乎放弃了视频生成模型的竞赛，认为 Google 对 Veo 的投入不足，而目前唯一持续进步的大型视频生成模型似乎只有 Seedance。
Andrej Karpathy discusses and breaks down the capability gap in a recent post on X, though the substance is contained in the linked tweet rather than the article itself.
An analysis of 412 humanoid robot job postings identifies the 5 robot companies best suited for career entry, highlighting Figure as the top hirer with 101 open roles and 63 requiring no specific educational background.
Multiple companies released coordinator tools for AI agents in the same week, but these tools scale existing code review processes without fundamentally changing the job.
This article critiques the exaggerated hype around recent AI events, such as breakthrough claims and security incidents, showing how expert scrutiny often reveals more subdued realities than the initial press coverage suggests.
A tweet discusses a slide from a talk at Sequoia’s AI Ascent, highlighting the early stage of AI application across industries and its large total addressable market.
The article speculates that AI companies may be nearing the end of significant LLM advancements, with rising training costs and diminishing returns potentially leading to slower progress to maintain hype.
The article argues that US safety regulations for AI might be partly motivated by OpenAI and Anthropic's declining market share, with anecdotal evidence of companies switching to more efficient alternatives like DeepSeek, and draws historical parallels to the 'war of the currents'.
The article explores the difference between internal and external views on AI progress, emphasizing the role of scaling laws and capability jumps in shaping perceptions of future advancements.
The AI buildout forecast to 2030 predicts $3.8 trillion in capital expenditure and job disruptions, but argues that accelerating development will yield greater societal benefits through abundant intelligence.
This article presents a list and analysis of top enterprise AI development companies for 2026, evaluated on criteria like customization, domain expertise, and integration readiness, to assist businesses in selecting partners.
Jensen Huang, CEO of Nvidia, explains why the company expects 70% revenue growth next year, citing its central role in the AI ecosystem and addressing competition concerns.
The article critiques the 'Security as Identity' mindset in the security industry, arguing that prioritizing hacker aesthetics over effective solutions leads to missed opportunities, drawing an analogy to environmentalism's historical failures.
The article argues that OpenAI's claimed 3x AI productivity gain comes from AI systems working continuously like extra shifts, rather than making humans more efficient, and highlights the high costs and defect rates involved.