According to reports citing informed sources, Anthropic, a leader in the generative AI field often dubbed the "OpenAI rival," projected during an internal meeting in December 2025 that it could achieve up to $18 billion in revenue for 2026. This represents an approximately 20% increase from its summer forecast, with expectations to reach roughly $55 billion the following year—indicating that Anthropic believes its revenue could more than triple within the next year, driven by its popular core AI application products like the AI programming tool Claude Code and the Sonnet series. Media reports indicate that in its most optimistic projections, Anthropic—backed by major tech giants Amazon (AMZN.US) and Alphabet's Google (GOOGL.US)—anticipates total revenue could reach as high as $148 billion by 2029, which is about $3 billion higher than OpenAI's projected revenue for the same year. Anthropic did not immediately respond to any media requests for comment. Informed sources stated that the vast majority of the rapidly growing revenue will come from Anthropic successfully selling access to its flagship AI applications via an Application Programming Interface (API) to a large base of enterprise clients. The company has become particularly popular for AI programming-related tasks executed by artificial intelligence agents, partly due to its agentic AI programming application, Claude Code; this product reportedly generated over $1 billion in annualized revenue back in November 2025. AI programming is arguably the hottest technical sub-sector within the global wave of AI applications. It is becoming an inevitable trend in both B2B and C2C markets, accelerating its penetration worldwide and evolving from a "personal developer tool" into an "enterprise-grade standard." The recent global popularity of AI programming highlights the arrival of a super-wave of AI applications sweeping across all industries, suggesting that epoch-making AI technology will undoubtedly lead a major productivity transformation comparable to the industrial revolution across various sectors. However, similar to cash-burning OpenAI, costs are also accumulating rapidly. Anthropic has revised its most optimistic forecast, pushing back the timeline for achieving positive cash flow to 2028—a full year later than previously expected. Anthropic anticipates generating $2.2 billion in cash during 2028. Media reports suggest that the higher costs associated with training and efficiently running its large AI models contributed to this delay. For 2026, Anthropic preliminarily expects to spend $12 billion on training models and approximately $7 billion on efficiently running the large AI model technology for its paying users. The AI unicorn is currently raising over $10 billion in a funding round led by Singapore's sovereign wealth fund GIC and Coatue Management, at a pre-money valuation of $350 billion. Last week, media reports indicated that Anthropic's latest annual revenue run-rate had exceeded $9 billion by the end of 2025. In November 2025, Nvidia and Microsoft reportedly committed to investing up to $10 billion and up to $5 billion in Anthropic, respectively. Microsoft has already invested over $13 billion in OpenAI, giving it a nearly 27% stake in the ChatGPT developer. Earlier this month, reports suggested that Microsoft has become one of Anthropic's top customers and is recently expected to purchase AI large model technology exclusively developed by Anthropic at a scale of approximately $500 million per year to support Microsoft's vast software product portfolio, highlighting that Microsoft is unwilling to place a 100% bet on OpenAI and prefers to diversify its investments across more leading AI application companies. Claude Code has recently exploded in popularity within the "AI programming/Agentic coding" community. It goes beyond merely "chatting to write code" by productizing the entire engineering workflow—"write, modify, test, run, fix, submit PR/deliver"—into a sustainable and reusable working method. One of the strongest signals is Anthropic's own disclosure: Claude Code achieved a $1 billion run-rate revenue level within about six months of its public release (a financial metric used to extrapolate annual total revenue based on current performance, often used for forecasting or evaluating business performance, particularly for new ventures or growing companies), which typically only occurs when developers and enterprise customers pay frequently and consistently. Claude Code is defined as an "agentic coding tool that lives in the terminal," naturally integrating with developers' daily surfaces like repositories, CLI, git, and CI, thereby reducing migration costs. The core capability of Claude Code is its "agentic" nature rather than "single-turn Q&A," which is a key factor in its rapid global popularity. Anthropic emphasized in its SWE-bench research that to translate model capabilities into real engineering output, it's necessary to build agents around the model (task decomposition, execution, verification, rollback, etc.), and Claude 3.5 Sonnet has delivered leading results on SWE-bench Verified, demonstrating its "actionable fixing/implementation capabilities" in both evaluation and practice. From a broader AI application trend perspective, programming is becoming the killer scenario where generative AI first achieves "quantifiable ROI": tasks are naturally verifiable (can compile/run/have tests), feedback loops are short, and it directly compresses delivery cycles. This is also why the personal AI assistant ClawdBot (recently renamed Moltbot) has recently sparked widespread discussion—it's not just an "executor" focused on the IDE/terminal but acts more like a cross-application personal/team coordination agent—initiating tasks, tracking status, and summarizing results through message/notification/lightweight command interfaces, even popularizing concepts like "vibe-coding." Its open-source and self-hostable nature further amplifies its diffusion speed. From a macro perspective on AI application trends, Claude Code + (ClawdBot/Moltbot) represent the same main theme: AI monetization is shifting from "selling models" to "selling workflows." The coding scenario is naturally verifiable (can compile/run/have tests), making it easier to first establish enterprise-level ROI; the next phase of competition will shift focus from "single-model capability" to "multi-agent orchestration, observability, permission openness, and security-first architecture," expanding from developer surfaces to more organizational surfaces (business systems, tickets, operations, customer service, etc.). AI programming is accelerating its penetration globally! The wave of AI applications centered around agentic workflows has arrived. For the bullish narrative logic of the global stock market's AI application sector, Anthropic's raised revenue expectations, as an "OpenAI rival," undoubtedly serve as a positive growth signal, "validating the feasibility of the AI application story + preheating the potential accelerated growth trend post-2026." AI programming is currently the hottest technical sub-sector within the global wave of AI applications. In both B2B and C2C markets, agentic "vibe coding" based on AI has become an inevitable trend and is accelerating its penetration worldwide. So-called "Vibe coding" can be understood as a more radical, colloquial expression for AI-assisted coding. It emphasizes using pure human natural language prompts to get large AI models to generate, modify, and debug code, with the human role shifting from "writing code" to "describing goals and providing iterative feedback"; some engineering practices even emphasize "getting it running first," reducing the requirement for line-by-line code understanding and review. Since 2025, cloud computing and search engine leader Google, AI application leader Applovin focused on "AI + digital advertising," AI software platform leader Palantir focused on "AI + data analysis," and "internet gatekeeper" Cloudflare with its "Connectivity Cloud" positioning have all announced exceptionally strong performance data and future outlooks. This indicates that not only is the demand for AI computing infrastructure, represented by Nvidia's AI GPUs, exceptionally strong, but the demand for AI software applications, especially enterprise-grade AI application software that can comprehensively enhance B2B operational efficiency, is also robust and accelerating its penetration into various industries. Based on the current technological trajectory, the development direction of AI application software is concentrated on "generative AI application software" (such as the globally popular DeepSeek, ChatGPT, Sora, and Anthropic's Claude), and, building on generative AI, the evolution of AI functionality from chatbox-style Q&A to "agentic AI intelligences that autonomously execute various tedious and complex tasks." Businesses' urgent need to improve efficiency and reduce operational costs has recently been a significant driver for the widespread application of the two core categories of AI application software—generative AI applications and AI agents. Among these, AI agents are highly likely to be the major AI application trend before 2030. The emergence of AI agents signifies that artificial intelligence is beginning to evolve from an information assistance tool into a highly intelligent productivity tool. The latest research from MarketsandMarkets shows that the AI agent market size is expected to reach as high as $53 billion by 2030, implying a Compound Annual Growth Rate (CAGR) of 46% starting from 2025.
Comments