Anthropic is preparing for its largest-ever IPO, building on a year where the company demonstrated that heavily capitalized AI firms can achieve profitable operations.
In the AI industry, Anthropic has become impossible to ignore. Over the past year, the company has demonstrated to the market that heavily capital-intensive AI ventures can indeed achieve sustainable profitability. While most AI firms remain anxious about boasting impressive technology but weak financials, Anthropic has delivered a performance report that has stunned the entire sector.
According to Bloomberg, as of late July 2026, Anthropic’s annualized recurring revenue (ARR) reached $65 billion, surging more than sevenfold from $9 billion at the end of 2025. With its expected October IPO window approaching, the company is targeting a valuation of $2 trillion, aiming to set a new global record for initial public offerings.
It is important to distinguish between annualized recurring revenue and actual booked annual revenue. The $65 billion figure does not represent fully realized yearly income; rather, it is a forward-looking projection based on current business velocity, making it a reliable metric for gauging the operational momentum of rapidly scaling enterprises.
Anthropic’s growth curve has been exceptionally steep. After hitting $9 billion in ARR at the end of 2025, it climbed to $47 billion by May. Within just two months, it rose another 38% to reach $65 billion. Meanwhile, the company’s actual second-quarter revenue totaled $11.5 billion, representing a year-over-year surge of approximately 14.6 times compared to $787 million in the same period last year.
More significantly, this frontier large-language-model company, known for its heavy compute investments, has achieved positive adjusted operating profit. For years, the industry operated under the assumption that even explosive revenue growth could never outpace the burning cash required for compute infrastructure, talent acquisition, and data centers. Anthropic has broken this spell. Turning a profit signals that enterprise deployment has moved beyond theoretical promise and is now delivering tangible operating leverage.
This shift is largely driven by Claude Code, which has catalyzed enterprise-grade agent workflows. Historically, large language models were primarily used for single-turn question-and-answer interactions, characterized by short call durations and limited token consumption. Claude Code, however, can read code repositories, modify files, execute commands, and handle debugging and iterative development, supporting long-running, continuous software engineering tasks.
Enterprises are no longer treating the model as a chatbot but are embedding it directly into real-world business processes, causing token consumption to multiply. Currently, 80% of the company’s revenue comes from API and enterprise clients under a usage-based pricing model, distinct from traditional seat-based subscriptions. As clients expand their operations, revenue grows accordingly, creating a self-reinforcing growth flywheel.
Bloomberg previously reported that OpenAI’s ARR stands at approximately $40 billion. On a direct comparison, Anthropic’s $65 billion ARR has already surpassed its rival.
Standing on the eve of its IPO, these compelling financial figures serve as Anthropic’s strongest leverage to the capital markets. Earlier reports from Britain’s Financial Times indicated that several investors anticipate the company will go public in October at a valuation of $2 trillion or higher. Institutional forecasts project that Anthropic’s revenue could reach $190 billion to $200 billion by 2028. If realized, this would surpass SpaceX to become the largest IPO in history.
It is evident that the capital markets’ evaluation criteria for AI companies are evolving. The focus is shifting away from model capabilities alone, with increasing weight placed on enterprise penetration rates, growth velocity, and profitability.
Beneath the highlights, risks persist. Regulatory uncertainty remains a constant overhang. In June, two of its flagship models were temporarily taken offline following U.S. export control policies. Although services were restored after negotiations, the ongoing tug-of-war between the company and regulators shows no signs of abating.
Anthropic co-founder Dario Amodei has publicly addressed AI’s inherent tendency toward centralization of power, advocating for a regulatory framework that balances risk management, curbs the dominance of top-tier giants, and preserves space for open-source models to thrive.
Looking ahead, Anthropic faces pressing practical questions: how to absorb escalating compute costs, whether its profitability trajectory can be sustained, and if its lofty valuation can hold firm. These remain the defining tests for the company.