Anthropic's Q2 earnings of 21.6 trillion won still resulted in a loss exceeding 10 billion won.

Anthropic achieved $16.3 billion in revenue in the first half of the year, representing a 1,500% surge compared to the previous year's same period. Despite this growth, the company recorded a net loss of $80.7 billion, which is nearly double the prior year's annual loss. The majority of the reported loss, $72 billion, is attributed to accounting fluctuations related to fundraising instruments like convertible bonds. This financial performance was driven by substantial investment in infrastructure, with $9 billion spent in the first half on computing capacity and infrastructure, exceeding the prior year's annual expenditure. Future resource contracts are projected to be immense, including obligations to major cloud providers and chip leasing agreements, leading to estimated future costs exceeding $600 billion. Furthermore, securing necessary computing capacity is estimated to cost up to $750 billion in total, considering ongoing negotiations. The company faces technological risks related to AI model control and potential liabilities from infrastructure accidents.

The narrative presents a stark contrast between explosive revenue growth driven by market perception and enormous operational expenditure required to support that growth. The core tension lies in how Anthropic leverages massive infrastructure spending—which functions as a barrier to entry—to secure future relevance while simultaneously absorbing significant losses. The astronomical projections for future computing costs, running into the trillions of won, suggest that the current financial structure is heavily predicated on ongoing, extremely high-stakes resource acquisition rather than sustainable profitability from service provision alone. The mention of competitors building their own data centers highlights a systemic risk: reliance on third parties creates dependence and exposes the firm to competitive dynamics where infrastructure costs dictate strategic positioning. The discussion of AI safety risks—specifically self-preservation properties and catastrophic accident potential—is juxtaposed against this relentless pursuit of computational scale, suggesting an underlying tension between unchecked technological ambition and fiduciary responsibility. The implications point toward a future where access to foundational computing capacity becomes the primary determinant of enterprise viability, raising questions about how value is measured when the cost of existence (computing) dwarfs revenue.