单词学习：![[2026031630628396.mp3]]

新闻播报：![[2026031678648974.mp3]]

奥尔特曼：AI将演变为基础公用事业按Token计费，行业面临巨大算力与基建挑战

简报：

• OpenAI CEO 山姆·奥尔特曼在贝莱德基础设施峰会上表示，未来人工智能将像水和电一样成为基础公用事业，模型提供商将按处理数据的单位（Token）的实际使用量进行销售和计费。
• 奥尔特曼指出，算力将决定AI的使用权，若无法构建足够算力满足需求，AI服务将变得极其昂贵并向富裕阶层倾斜，或需政府介入分配。
• 为应对飙升的AI需求，科技公司今年将在算力领域投入数千亿美元；OpenAI总裁格雷格·布罗克曼此前透露，公司已承诺未来八年投入约1.4万亿美元用于数据中心项目。
• 算力需求呈指数级增长，AMD CEO苏姿丰在2026年CES上预测，未来五年全球需超10 yottaflops的算力，达到2022年规模的1万倍。
• 算力扩张正面临耗电量大及电网压力等基建瓶颈，埃隆·马斯克此前预测，得益于更快的能源基础设施建设速度，中国的人工智能总算力可能会超越美国。

相关链接：

• OpenAI CEO 奥尔特曼：未来 AI 将如水电般按量计费

Altman: AI Will Evolve into a Basic Utility Billed by Token, as the Industry Faces Massive Computing and Infrastructure Challenges ⚡🤖

Brief:

• OpenAI CEO Sam Altman said at BlackRock’s Infrastructure Summit that artificial intelligence will eventually become a basic utility like water and electricity, with model providers selling and billing based on actual usage measured in units of processed data, or tokens.
• Altman noted that computing power will determine access to AI. If enough capacity cannot be built to meet demand, AI services could become extremely expensive and skew toward the wealthy, potentially requiring government intervention in allocation.
• To cope with surging AI demand, tech companies are set to invest hundreds of billions of dollars in computing this year. OpenAI President Greg Brockman previously revealed that the company has committed about dJyH6kDkfC5RlM16.answerText.4 trillion over the next eight years to data center projects.
• Demand for computing power is growing exponentially. At CES 2026, AMD CEO Lisa Su predicted that the world will require more than 10 yottaflops of computing power over the next five years—10,000 times the 2022 level.
• The expansion of computing capacity is facing infrastructure bottlenecks, including enormous electricity consumption and pressure on power grids. Elon Musk previously predicted that, thanks to faster energy infrastructure development, China’s total AI computing capacity could surpass that of the United States. 🌍

Related link:

• OpenAI CEO Altman: In the future, AI will be billed like water and electricity based on usage

特斯拉加速“造芯”自主化，AI芯片巨型工厂项目即将启动

简报：

• 马斯克于3月14日正式宣布，特斯拉专门用于生产人工智能芯片的巨型工厂（Terafab）项目将在7天后正式启动。
• 建设该工厂旨在为FSD（全自动驾驶）提供算力底座，以解决外部供应商产能无法满足特斯拉AI芯片需求的问题，推动实现从设计到制造的全链路自主化。
• 目前特斯拉正在全力设计第五代人工智能芯片（AI5），在推进自建工厂的同时，仍保持与台积电、三星等外部代工厂的合作。
• 因传闻三星2纳米工艺流片延期，特斯拉下一代AI6芯片的量产时间已被迫推迟至2027年底，自建芯片工厂项目将有助于特斯拉化解外部供应链风险。

相关链接：

• 马斯克下令特斯拉工厂下周启动AI芯片制造项目

Tesla Speeds Up In-House Chip Manufacturing, AI Chip Gigafactory Project Set to Launch 🚀

Brief:

• On March 14, Elon Musk officially announced that Tesla’s gigafactory project dedicated to producing artificial intelligence chips, known as Terafab, will officially begin in seven days.
• The factory is being built to provide the computing foundation for FSD (Full Self-Driving), addressing the issue that external suppliers’ production capacity cannot meet Tesla’s AI chip demand, and helping advance full-stack independence from design to manufacturing.
• Tesla is currently pushing ahead at full speed with the design of its fifth-generation artificial intelligence chip (AI5). While promoting the construction of its own factory, the company is still maintaining partnerships with external foundries such as TSMC and Samsung.
• Due to rumors that Samsung’s 2-nanometer process tape-out has been delayed, mass production of Tesla’s next-generation AI6 chip has reportedly been pushed back to the end of 2027. Tesla’s self-built chip factory project is expected to help the company mitigate external supply chain risks. 🤖

Related link:

• Musk Orders Tesla Factory to Launch AI Chip Manufacturing Project Next Week 🔗

Google地图迎十年来最大升级：接入Gemini推出AI智能助理与3D实景导航

简报：

• 谷歌CEO宣布Google地图接入Gemini模型，推出“Ask Maps”与“沉浸式导航”两大功能，官方称此为该应用十多年来最重大的一次升级。
• “Ask Maps”支持用户通过自然语言进行复杂条件提问，系统会分析海量评论、照片和实时流量数据提供定制建议，并支持直接订位。
• “沉浸式导航”利用Gemini对数亿张街景和航拍照片进行实时渲染，呈现包含建筑物、绿植、交通信号灯及人行横道等微小细节的3D上帝视角。
• 智能化体验升级还包括：即将到达目的地时主动高亮显示建筑入口和附近停车位，以及通过分析全球每秒500万条交通信息来对比路线优劣。
• 目前“Ask Maps”功能已率先在美国和印度上线，“沉浸式导航”也将陆续覆盖iOS和Android设备。

相关链接：

• Google地图接入Gemini：支持一句话生成攻略及3D沉浸式导航

Google Maps Gets Its Biggest Upgrade in a Decade: Gemini-Powered AI Assistant and 3D Immersive Navigation 🚗✨

Brief:

• Google CEO announced that Google Maps is integrating the Gemini model and launching two major features: “Ask Maps” and “Immersive Navigation.” Officials describe this as the most significant upgrade to the app in more than a decade.
• “Ask Maps” allows users to make complex queries in natural language. The system analyzes massive volumes of reviews, photos, and real-time traffic data to provide personalized recommendations, and also supports direct reservations.
• “Immersive Navigation” uses Gemini to render hundreds of millions of Street View and aerial images in real time, creating a 3D bird’s-eye view with fine details such as buildings, greenery, traffic lights, and crosswalks. 🗺️
• Other smart experience upgrades include proactively highlighting building entrances and nearby parking spaces as users approach their destinations, as well as comparing route options by analyzing 5 million traffic updates per second from around the world.
• Currently, the “Ask Maps” feature has already launched in the United States and India, while “Immersive Navigation” will gradually roll out to iOS and Android devices.

Related link:

• Google Maps Integrates Gemini: Supports One-Sentence Trip Planning and 3D Immersive Navigation

英伟达CEO黄仁勋于GTC 2026大会重磅展示硬件技术新品NemoClaw

简报：

• 英伟达CEO黄仁勋在GTC大会上重磅展示了最新研发的硬件与技术新品NemoClaw
• 此次新品展示再次向业界秀出了英伟达在AI算力与前沿生态上的实力与肌肉

相关链接：

• 黄仁勋在GTC 2026展示辉达新品NemoClaw

NVIDIA CEO Jensen Huang Unveils Major New Hardware Technology Product NemoClaw at GTC 2026 🚀

Briefing:

• NVIDIA CEO Jensen Huang made a major appearance at the GTC conference to unveil NemoClaw, the company’s latest hardware and technology product 🤖
• The new product showcase once again demonstrated NVIDIA’s strength and dominance in AI computing power and its cutting-edge ecosystem 💪

Related Link:

• Jensen Huang Unveils NVIDIA’s New Product NemoClaw at GTC 2026

腾讯云大模型最高涨价逾400%，全球云厂商集体调价加剧AI阶层分化

简报：

• 腾讯云调整智能体开发平台计费策略，GLM 5、MiniMax 2.5等模型结束免费公测步入商用，混元系列部分模型价格涨幅超过400%。
• 亚马逊AWS、谷歌云以及国内的优刻得等全球云厂商此前已纷纷上调服务价格，打破了云计算行业二十年来“只降不涨”的常态。
• 调价的现实原因在于2026年初企业级AI应用的爆发导致Token消耗海量增长，旧硬件处理成倍推理需求导致成本承压，云厂商借此进行“现金流自救”。
• 尽管专用推理芯片使单Token物理成本下降，但复杂AI任务诱发了更高的调用频率，导致总账单攀升。
• 算力成本的飙升促使AI智能走向“阶层化”：资金雄厚的巨头可应用顶尖Agent实现生产力跨越，而小微企业和普通人可能被限制在低配版AI中，技术鸿沟进一步拉大。

相关链接：

• 腾讯云调价终结大模型降价潮，AI产业显现阶层化趋势

Tencent Cloud’s Large Model Prices Surge by Over 400%, as Global Cloud Providers Raise Prices and Deepen AI Stratification 🤖☁️

Briefing:

• Tencent Cloud has adjusted the pricing strategy of its agent development platform. Models such as GLM 5 and MiniMax 2.5 have ended their free public beta period and entered commercial use, while prices for some models in the Hunyuan series have increased by more than 400%.
• Global cloud providers, including Amazon AWS, Google Cloud, and domestic firms such as UCloud, had already raised service prices earlier, breaking the cloud computing industry’s two-decade norm of “prices only go down, never up.”
• The practical reason behind these price hikes is the explosive growth of enterprise AI applications expected in early 2026, which has driven massive increases in token consumption. Older hardware is struggling to handle multiplying inference demand, putting heavy pressure on costs and prompting cloud providers to pursue “cash flow self-rescue.”
• Although dedicated inference chips have reduced the physical cost per token, more complex AI tasks have triggered higher call frequency, causing total bills to rise 📈.
• The sharp increase in computing power costs is pushing AI intelligence toward “stratification”: well-funded giants can deploy top-tier agents to achieve major productivity gains, while small businesses and ordinary users may be confined to lower-tier AI versions, further widening the technological gap.

Related link:

• Tencent Cloud Price Hikes End the Downward Trend in Large Model Pricing, as AI Industry Shows Signs of Stratification