AI两大头部企业的盈利路径正面临严峻考验。 据The Information报道，OpenAI与Anthropic均未能达成各自设定的毛利率目标，推理成本的超预期增 长是主要拖累因素。这一财务压力在用户规模持续扩张的背景下尤为突出，引发外界对两家公司能否在 本十年末实现60%以上毛利率目标的质疑。 数据显示，OpenAI去年毛利率从40%降至33%，低于其自身预测的46%；Anthropic方面，毛利率从2024 年的负94%大幅改善，但预计2025年毛利率为40%，仍较此前目标低出10个百分点。两家公司目前均未 面临融资困境，但投资者对其商业模式的可持续性正给予更多审视。 推理成本失控，超出自身预测 OpenAI的毛利率压力还来自其庞大的非付费用户群体。据The Information，OpenAI每周活跃用户约达 9.1亿，其中付费用户占比仅约5%。去年，其总推理成本中近半数（约39亿美元）用于支撑非付费用 户，付费用户对应成本为45亿美元。 产品结构同样是一大因素。OpenAI去年推出的视频生成工具Sora对服务器算力的消耗远高于文本类查 询；其推理模型在计算答案时所需算力也高于传统大语言模型。此外，该 ...

Fears over artificial intelligence are now hitting bond investors, according to the latest credit investors survey from Bank of America Global Research.An AI bubble is seen as the top risk for the first time, with investors expecting a surge in bond issuance from hyperscalers, the February poll found. Some 23% of those surveyed said it was their biggest concern, compared to 9% who said so in the firm's December poll. Companies including Amazon, Google parent Alphabet and Meta have announced big spending pla ...

People are calling anthropic hypocrites, including Elon Musk. Here's the post, by the way. 26 million views. We've identified industrialcale distillation attacks on our models by Deepseek, Moonshot AI, and Mini Max. These labs have created over 24,000 fraudulent accounts and generated over 16 million exchanges with Claude, extracting its capabilities to train and improve their own models. Foreign labs that illicitly distill American models can remove safeguards. feeding model capabilities into their own mil ...

来源：环球市场播报 软件板块延续前一日涨势，iShares拓展技术与软件板块ETF（IGV）上涨约2%。该基金周三盘前微涨。 Palantir Technologies和微软等个股走高。但Workday因发布疲软营收预测大跌9%。 周二主要股指普遍上涨，因市场对人工智能颠覆多个行业的担忧有所缓解。 北京时间2月25日晚，美股周三高开。受英伟达和甲骨文股价上涨支撑，主要股指延续了前一交易日的 涨势。市场关注英伟达、赛富时及Snowflake等公司即将公布的财报。 英伟达、赛富时和Snowflake等将于周三收盘后公布业绩。当前投资者正在重新评估高企的科技股估 值，并对超大规模企业高额的人工智能资本支出产生疑虑。 英伟达近月表现逊于芯片行业的同类企业，这一趋势与纳斯达克100指数自去年10月底以来未能再创新 高、以及资金从科技股撤出的轮动现象相吻合。 为重振股价表现，英伟达至少需要超越此前的展望，并设定高于华尔街当前预估的新目标。尽管该公司 屡次做到了这一点，但市场对人工智能支出浪潮不可持续的担忧日益加剧。 瑞银全球股票首席投资官Ulrike Hoffmann-Burchardi在报告中表示："未来几日市场信心能 ...

Hut 8 (NasdaqGS:HUT) Q4 2025 Earnings call February 25, 2026 08:30 AM ET Company ParticipantsAsher Genoot - CEOSean Glennan - CFOSue Ennis - Head of Investor Relations and Government AffairsSue EnnisGood morning, welcome to Hut 8's full year 2025 financial results conference call. Joining us today are our CEO, Asher Genoot, and our CFO, Sean Glennan. Following the presentation, we will open the line for questions. This event is being recorded, and a transcript will be made available on our website. In addit ...

Anthropic on Tuesday announced a new suite of enterprise offerings, the latest in a series of new capabilities boasted by AI labs like Google, Anthropic, and OpenAI that have shaken investor confidence in a variety of software solutions that may see their core product replaced by an AI facsimile. But some Wall Street analysts aren't so sure the process will be quite that simple. Dan Ives, high profile tech bull at Wedbush, wrote in a note Tuesday after Anthropic's announcement that, "While these use case ...

Extensive peering and network path optimization via NewEdge Route Control enable a fast, seamless digital experience for business-critical AI applicationsSANTA CLARA, Calif., Feb. 25, 2026 (GLOBE NEWSWIRE) -- Netskope (NASDAQ: NTSK), a leader in modern security and networking for the cloud and AI era, today announced NewEdge AI Fast Path, a set of capabilities that efficiently optimize network paths to critical AI destinations, including AI applications hosted in public, private, or neo-clouds. Available no ...

Core Insights - The emergence of RentAHuman.ai highlights a new trend in the gig economy where individuals can be hired for various tasks, including those that AI cannot perform directly [1][3][4] - The platform has gained significant traction, with over 560,000 registered users and a task completion rate of 92.7% as of February 2026 [4][12] - The average hourly wage for human workers on the platform is $68, which is 51.1% higher than traditional gig platforms [4] Business Model - RentAHuman.ai operates similarly to a company that hires employees, utilizing initial funding of $12 million from investors like Multicoin Capital to pay human workers [12][10] - The platform generates revenue by outsourcing tasks from businesses that prefer not to incur direct labor costs, with 65% of its income coming from corporate clients [15][12] - The API subscription model allows other businesses to access human resources for a fee, contributing to an additional revenue stream of over $120,000 per month from 800 subscribers [19][17] Efficiency and Performance - The platform boasts a task matching efficiency of 2.8 minutes with a success rate of 96.3%, significantly outperforming traditional methods [36][34] - AI management reduces decision-making errors to 3.2%, compared to 27.8% in human management, indicating a substantial advantage in standardized tasks [43][42] - However, human oversight is still necessary to mitigate issues like fraud, as demonstrated by a previous incident where over 120 workers exploited AI's limitations [50][49] Future Outlook - Predictions suggest that within 24 months, the first "zero-employee company" could emerge, distributing over $100 million in wages to human workers [27][26] - By 2030, over 50% of standardized tasks may be managed by AI, with a projected 1 million AI employers hiring over 500 million human workers by 2035 [53][54] - The model of AI hiring humans is seen as complementary rather than substitutive, as AI currently lacks the physical capabilities to perform many tasks independently [70][69]

Ira Bodnar, founder of San Francisco–based startup Ryze, said her company's core product was effectively made obsolete overnight following rapid AI breakthroughs by Anthropic and Manus AI.Despite having gained several hundred paying clients in just two months, Ryze found its specialized category redundant due to the AI’s native automation.However, Bodnar remains optimistic about the future, announcing plans to pivot Ryze towards building complex AI workflows for large advertising agencies. Predictions On AI ...

Investment Rating - The report maintains a "Positive" investment rating for the industry [2] Core Insights - The industry is witnessing a threefold evolution in overseas large models, focusing on reasoning foundation, action implementation, and ecological reconstruction, with major players like Anthropic, OpenAI, Google, and OpenClaw leading the charge [7] - Domestic large model companies are breaking through through open-source foundations, efficiency optimization, and collaborative agent ecosystems, with firms like Z.ai, MiniMax, Kimi, Alibaba, and ByteDance showcasing diverse strategies [7] - The industry trend indicates a shift from generation to action, with competition centering on supply efficiency and ecological reconstruction, emphasizing task complexity and ROI as key growth drivers [7] Summary by Sections Overseas Large Model Evolution - Anthropic has completed a dual-version model iteration, focusing on enterprise workflows and enhancing coding and long-range agent capabilities [11] - OpenAI is concentrating on long-range task closure and tool execution, evolving its models to become collaborative productivity assistants [25] - Google has released Gemini 3.1 Pro to enhance reasoning capabilities and Lyria 3 to expand into audio creation, reinforcing its competitive edge [32][40] - OpenClaw is positioned as a self-hosted gateway, integrating multiple communication channels and supporting tool-based agents [43] Domestic Large Model Breakthroughs - Z.ai's GLM-5 aims to extend open-source model capabilities to complex systems and long-range agent tasks, with a focus on engineering deployment [48] - MiniMax's M2.5 emphasizes real-world productivity, optimizing costs and throughput to facilitate agent scalability [53] - Kimi's K2.5 leverages a multi-modal MoE architecture to enhance visual understanding and parallel agent execution [65] - Alibaba's Qwen3.5-Plus focuses on open-source upgrades and multi-modal transitions to drive agent scalability [73] - ByteDance's recent model releases aim to transition AI capabilities from dialogue to actual task execution [81] Industry Trends - The industry is transitioning from generation to action, with a focus on supply efficiency and ecological reconstruction, as evidenced by the increase in token processing volumes and the emergence of efficiency-friendly models [89]