Welcome to the Vergecast, the flagship podcast of putting as many megapixels as you possibly can into your selfie camera. I'm your friend David Pierce and I am slowly losing my mind if I'm being completely honest with you. So, like I mentioned last week on the show, I am shopping for a new phone.This is a very fun thing for me to do. For a long time, I was a phone reviewer. So, I'd lived a totally abnormal phone life.I would switch phones like every six weeks based on what I was reviewing. But for the last ...

【导读】Anthropic刚刚扔出一份18页重磅炸弹：《2026年智能体编码趋势报告》。结论直接炸裂：程序员不再写代码了，他们变成了「指挥官」。单个 AI助手已经进化成自主智能体军团，能花好几天打造完整系统，甚至让法务、市场这种完全不懂代码的人，也能自己做应用。软件开发，正在经历图形 界面发明以来最大的一次地震。 2026年的AI圈，要说谁最风光，Anthropic绝对排第一。 就在刚刚，Anthropic甩出了一份18页重磅报告：《2026年智能体编码趋势报告》。 这份报告的核心结论可以浓缩为一句话：任何人，都成为了开发者。 编程/软件开发的游戏规则彻底变了！ 注意，不是「每个程序员变得更强了」，而是「非技术人员也能开发了」。 这意味着软件开发这个行当，正在经历自DOS、图形界面发明以来，最大的一次范式转移。 以下是报告全部8大趋势的深度解读，干货满满，一个不漏。 写在前面 这份18页的报告，信息密度极高！ Anthropic没有画饼，没有讲故事，用的是Anthropic自己的内部研究数据、真实客户案例和对行业趋势的冷静判断。 核心信号非常清晰：程序员不会消失，但「只会写代码」的程序员会消失。 未来的软件工 ...

Core Insights - OpenAI has officially announced the testing of an advertising feature in the free and Go versions of ChatGPT across the United States, aiming to support free users due to financial constraints [1][5][6] Group 1: Advertising Implementation - The advertising feature will only be available to users of the free and Go versions, with no impact on other subscription tiers [6] - Ads will be clearly marked as "sponsored" and will not affect the quality or independence of ChatGPT's responses [6] - The ads shown will be determined by user chat history, current conversation topics, and interaction with previous ads, ensuring relevance [6][9] Group 2: Rationale Behind Advertising - OpenAI's motivation for introducing ads is to achieve widespread access to AGI, leveraging a proven business model that has worked for companies like Google and Facebook [5][7] - The introduction of ads is seen as a way to enhance user conversion rates, particularly for low-spending users, similar to Netflix's recent ad-supported subscription model [7] Group 3: Future of Advertising in AI - OpenAI aims to set higher industry standards for AI advertising, ensuring that the model remains unaware of ad content unless explicitly requested by the user [9] - The company believes that AI advertising will benefit small businesses by allowing them to create targeted marketing strategies through AI [9] - OpenAI envisions a future where ads evolve into more proactive interactions, aligning with user preferences and enhancing the overall experience [9] Group 4: Market Position and Growth - OpenAI has recently launched a new programming model, GPT-5.3-Codex, which has received positive feedback and increased user engagement by approximately 50% [10] - The monthly growth rate of ChatGPT has reportedly returned to over 10%, with a new version of the chat model expected to be released soon [10] Group 5: Funding and Industry Impact - OpenAI's $100 billion funding plan is gaining momentum, with the first part expected to come from major investors like Microsoft, NVIDIA, and Amazon, totaling up to $50 billion [12] - The successful integration of the "AI + advertising" model could trigger a ripple effect in the large model industry, prompting other companies to follow suit [12]

Group 1 - US stock index futures showed slight increases, with Nasdaq futures up 0.22%, S&P 500 futures up 0.15%, and Dow futures up 0.01% [1] - European stock indices had mixed results, with Germany's DAX up 0.03%, UK's FTSE 100 down 0.34%, France's CAC up 0.25%, and the Euro Stoxx 50 up 0.19% [1] - Citigroup warned that "AI disruption" could lead software stocks to retest their 2023 lows, with a potential long-term terminal value evaporation of about one-third [1] - AI startup Anthropic plans to secure 10 gigawatts of data center computing power, surpassing OpenAI [1] - The US plans to exempt companies like Amazon, Google, and Microsoft from upcoming chip tariffs, potentially linked to TSMC's investment commitments [1] - TSMC reported a 36.8% year-on-year increase in January sales to a record high of NT$401.26 billion, with a board-approved capital budget of $44.962 billion [1] - Amazon plans to launch a content trading platform allowing publishers to sell content to AI companies [1] - Alphabet's Waymo achieved fully autonomous driving operations in Nashville, Tennessee [1] - Tesla's North American sales head Raj Jegannathan has left the company amid challenges of declining global demand [1] - Salesforce reportedly laid off fewer than 1,000 employees earlier this month, affecting marketing and product management roles [1] Group 2 - Eli Lilly will acquire Orna Therapeutics for $2.4 billion to strengthen its position in the autoimmune sector [2] - Baidu is reportedly secretly launching the "O Plan," which insiders say is related to the Baidu app [2] - Qianwen APP is testing integration with Damai, covering the entire process from film recommendations to payment and ticket issuance [2] - AstraZeneca expects a 9% increase in product sales to over $55.5 billion by 2025, anticipating continued growth this year [2] - Coca-Cola's Q4 net revenue of $11.8 billion fell short of expectations, with an anticipated adjusted organic revenue growth of 4-5% this year [2] - BP's Q4 underlying replacement cost profit increased by 32% year-on-year to $1.54 billion, with stock buybacks paused [2] - Barclays reported a pre-tax profit of £1.9 billion in Q4, exceeding expectations, and committed to returning at least £15 billion to shareholders by 2028 [2] - ON Semiconductor's Q4 revenue declined by 11% year-on-year to approximately $1.53 billion, with first-quarter guidance falling short of expectations [2] - Ferrari's Q4 net profit of €381 million exceeded expectations, with orders already booked through the end of 2027 [2]

分组1 - BP reported Q4 revenue of $47.38 billion, a 3.6% year-over-year increase, exceeding estimates by $5.05 billion [2] - BP's Q4 Non-GAAP EPADS was $0.60, in line with expectations [2] - For 2026, BP anticipates reported upstream production to be slightly lower and underlying upstream production to remain broadly flat compared to 2025 [2] - BP has decided to suspend share buybacks and allocate excess cash to strengthen its balance sheet, aiming to invest in oil and gas opportunities [2] 分组2 - The Trump administration plans to provide tariff carve-outs for companies like Amazon, Google, and Microsoft, linked to investment commitments from TSMC [2] - TSMC has pledged to invest $165 billion in building capacity in the U.S. as part of this tariff exemption plan [2] - The administration's plans are still in flux and have not yet been finalized [2] 分组3 - Instacart has partnered with 1-800-Flowers.com, allowing U.S. customers to order flowers directly through the Instacart app [2] - This partnership aims to expand Instacart's offerings in time for Valentine's Day, a significant gifting holiday [2] - Instacart reported a 1.47% increase in shares, while 1-800-Flowers.com saw a 0.25% rise [2]

Core Insights - The competition between Anthropic and OpenAI has intensified with the launch of Claude Opus 4.6 and GPT-5.3-Codex, marking a significant phase in the practical application of large models [1] - The models are designed to enhance autonomous operational capabilities, addressing the commercial viability and user expectations of large models [1] Model Performance - In the Terminal-Bench 2.0 evaluation, Claude Opus 4.6 achieved a score of 65.4%, while GPT-5.3-Codex reached 77.3%, claiming the best coding performance [1] - Feeling AI's CodeBrain-1, based on GPT-5.3-Codex, ranked second globally with a score of 72.9%, making it the only Chinese team in the top 10 [2][3] CodeBrain-1 Features - CodeBrain-1 focuses on efficiently completing coding tasks by utilizing useful context and reducing noise, which helps mitigate the hallucination issues of large language models [9] - It employs a validation feedback mechanism that allows it to learn from errors, thus shortening the generate-validate cycle [9][10] - The model dynamically adjusts plans and strategies, enhancing its operational success rate in real terminal environments [10][11] Terminal-Bench 2.0 Overview - Terminal-Bench 2.0, developed by Stanford University and Laude Institute, is a rigorous benchmark for evaluating AI agents in real command-line environments, with tasks that are complex and require multi-step solutions [13][17] - The benchmark's high difficulty level means that even top models typically score below 65%, highlighting the challenges AI faces in complex system-level tasks [17] Strategic Implications - The emergence of CodeBrain-1 signifies a shift towards a more dynamic interaction model in AI, where the focus is on workflow and application rather than just model capabilities [18] - The competitive landscape is evolving, with Chinese teams like Feeling AI positioning themselves as framework definers in the AI technology innovation path [19]

知情人士透露，华尔街大型私人信贷公司之一阿波罗全球管理公司（Apollo Global Management） 即 将达成一笔交易：向一家投资工具提供约34 亿美元贷款，用于购买英伟达芯片，并租赁给刚与 SpaceX 合并的埃隆・马斯克旗下 xAI。 这笔交易将是阿波罗第二次大手笔投资用于向 xAI 租赁芯片的投资工具。去年 11 月，阿波罗已提供过 一笔类似的35 亿美元贷款。该人士称，阿波罗还计划对新设立的投资工具进行股权投资，该工具目标 是合计募集53 亿美元的股权与债务资金。 这笔投资最快将于本周敲定，目的是缓解 xAI 的部分资金压力。马斯克正雄心勃勃地打造全球最大规 模的数据中心，用于训练 AI 模型。长期投资马斯克旗下公司的Valor Equity Partners 正在安排此次交 易。 该投资工具是 Valor 更大规模计划的一部分：计划募集200 亿美元的股权与债务，用于采购 AI 芯片并部 署在 xAI 数据中心。 过去，包括格拉西亚斯在内的 Valor 合伙人曾在关键时期进驻马斯克旗下公司提供支持：包括特斯拉产 能危机期间、以及马斯克重组推特（后更名为 X）之后。 管理资产规模超9000 ...

Core Insights - Companies are experiencing transformational challenges due to the rapid evolution of AI agents, which necessitates strategic adjustments to capture their value [3][4] - The development and scaling of generative AI use cases are complex, leading to hesitance among executives regarding immediate investments [3][4] Group 1: Key Trends Driving AI and Agent Development - AI agents are becoming increasingly capable of executing tasks and interacting with humans, lowering the barriers to AI application and indicating a potential reshaping of business processes [4] - The number of advanced language models has grown significantly, with an annual increase of 167% since 2020, and the success rate of AI agents completing long tasks has doubled approximately every seven months [8] - Investment in AI training has surged, with major cloud service providers planning to invest over $250 billion in AI and data centers by 2025 [8] Group 2: Strategies for CEOs to Capture AI Value - CEOs must fundamentally rethink operational methods, innovation mechanisms, and value propositions to harness the advantages of AI agents [6] - Key strategies include accelerating innovation, embedding AI into workflows, and fostering a culture of continuous learning and adaptation [9][10] - Early implementations of AI agents have shown significant value, such as reducing project cycles by 40-50% and costs by over 40% [10] Group 3: Organizational Transformation and AI Integration - Companies need to transition from viewing AI agents as mere tools to recognizing them as complex systems capable of executing intricate tasks [10] - The integration of AI agents into existing workflows requires careful planning and governance to avoid operational chaos and ensure alignment with business objectives [26] - A shift towards a "smart agent first" approach is essential for redesigning workflows and operational models, particularly in cross-functional processes [15][16] Group 4: Implementation Roadmap for AI Transformation - A two to three-year roadmap is proposed for CEOs to guide their organizations through the AI transformation journey, focusing on key milestones and decisions [17][18] - The first year should concentrate on building a unified understanding and laying the groundwork for scaling AI operations, with efficiency improvement targets set at 10% [19] - In the second and third years, the focus should shift to scaling successful AI implementations and rethinking business models to leverage AI's full potential [24][25] Group 5: Talent and Workforce Management in the AI Era - The workforce will need to adapt to new roles that involve managing and supervising AI agents, necessitating a shift in training and performance evaluation systems [23] - Companies should aim for 25-50% of employees to regularly use AI tools, integrating these capabilities into daily operations [21] - As AI agents take on more tasks, the demand for certain job roles will decrease, requiring strategic workforce planning and reskilling initiatives [25][26]

Core Insights - OpenAI has begun testing an advertising feature for ChatGPT in the U.S., marking a significant step in its commercialization efforts [1] - The initiative aims to generate advertising revenue to cover infrastructure costs while maintaining free service for users [1][2] - The move has sparked widespread controversy and backlash from users on social media [1][4] Commercialization Path - The advertising test is based on the principle of "task consistency," allowing OpenAI to sustain its free and low-cost service tiers [2] - Ads will be displayed based on user conversation topics and past interactions, similar to traditional search engines [2] - Users can opt for an ad-free experience by upgrading to a premium subscription or manage ad personalization settings [2] Privacy Commitments and User Control - OpenAI has outlined data protection measures, ensuring advertisers cannot access user chat histories or personal details [3] - Ads targeting users under 18 will be automatically blocked, and sensitive topics will not feature ads [3] - Users have control over ad preferences, including the ability to reject specific ads and manage personalized settings [3] Market Reaction and Competitive Dynamics - The introduction of ads has led to negative feedback on social media, with users expressing disappointment and concerns about the impact on ChatGPT's utility [4][5] - Competitor Anthropic has capitalized on this sentiment by mocking OpenAI's ad strategy in a Super Bowl advertisement [5] - Some users have expressed understanding, provided that ads do not interfere with the quality of generated responses [5] Financing Prospects and Product Iteration - OpenAI is accelerating product iterations alongside the ad testing, with positive reception for its new programming model, GPT-5.3-Codex [6] - The company is pursuing a massive financing round, potentially reaching $100 billion, to support its operations and prove the viability of its business model [7] - Successful implementation of the "AI + advertising" model could lead to a ripple effect in the industry, prompting other companies to accelerate their commercialization efforts [7]

Core Insights - OpenAI has announced the introduction of advertisements in its ChatGPT platform, which will appear at the bottom of responses for free and Go subscription users, while Plus, Pro, business, and educational users will not see ads [1][10] - The company aims to use ad revenue to support the infrastructure and operational costs of free and low-cost services, ensuring broader access to AI technology [10][12] - OpenAI emphasizes that ads will not affect the quality of responses and will be clearly distinguished from the content generated by the AI [1][5] Advertising Implementation - Ads will be tested primarily for free and Go subscription users, with a monthly fee of at least $20 required to avoid ads entirely [1][10] - OpenAI's CEO Sam Altman indicated that the monthly growth rate of ChatGPT has returned to over 10% and that an updated chat model will be deployed soon [3] - The company has established high standards for advertising, ensuring that ads are relevant, useful, and do not compromise user privacy [5][6] User Experience and Privacy - OpenAI has set principles to maintain user trust, including keeping ads separate from AI responses, ensuring privacy in conversations, and allowing users to control ad personalization [5][7] - Users will have the option to provide feedback on ads and can disable personalized ad settings, ensuring that sensitive topics are not targeted by ads [10][12] - The company aims to balance profitability with user experience, learning from early user feedback during the testing phase [10][12] Competitive Landscape - Other companies, such as Adobe and Target, have begun to announce their participation in the ad testing phase, indicating a potential shift in the AI industry's approach to monetization [12][13] - Competitors like Anthropic have committed to not using ads, highlighting differing business models within the AI sector [12][13] - OpenAI's approach may influence how other AI companies consider integrating advertising into their platforms in the future [12][13]