Understanding the Costs of AI: The Anthropic Dilemma
The question of how much AI costs has become a hot topic, especially highlighted by Anthropic’s staggering expenses this year. As reported, AWS has billed Anthropic a jaw-dropping $2.66 billion in 2023 alone. What’s critical to note, however, is that Anthropic has reportedly earned $2.55 billion during the same period, meaning it has spent more than it earned. This financial imbalance raises significant concerns about the sustainability of AI startups.
The Significant Financial Challenges Faced by AI Startups
The data revealed by Ed Zitron sheds light on a pressing issue that all AI startups face: they are spending significantly more than they are earning. According to Zitron, Anthropic’s earnings in 2024 were estimated to be between $400 million and $600 million, juxtaposed against its massive spending of $1.35 billion on AWS. This resulted in an overwhelming expenditure rate of 226% of its income. The outlook for 2025 doesn’t seem promising either, as expenses on AWS are predicted to consume 104% of revenue.
The reliance on AWS is only one part of the equation. Notably, Anthropic also utilizes Google Cloud infrastructure, likely compounding its operational costs elsewhere. This situation hints at the potential for crippling expenses that many startups are grappling with.
The Mystery of Unaccounted Expenses
Calculating Anthropic’s total spending reveals astonishing gaps in unexplained costs. In 2024, Anthropic’s overall spending was estimated to be $6.2 billion. With $1.35 billion accounted for in AWS fees, this leaves $4.85 billion of expenses that remain unaccounted for. What could these expenditures entail? Speculations point towards astronomical costs related to Google Cloud and other operational necessities, underlining that computing costs could be much higher than publicly understood.
Desperation for Continued Investment
Despite these financial trials, Anthropic has continued to seek capital. Zitron’s analysis notes that between 2023 and 2025, the company managed to engage in investment rounds totaling an astonishing $37.5 billion, including $20 billion in 2025 alone. Notably, a significant portion of the funding has come from infrastructure providers like Amazon and Google, creating a cycle of reliance on the very entities representing impending financial burdens.
Yet, despite this influx of capital, Anthropic appears as desperate for additional funding as its competitor, OpenAI. Strategies to acquire funds have even involved reaching out to nations in the Middle East, suggesting a precarious existence hinged on continued financial backing.
Escalating Costs Over Time
The rates at which Anthropic intends to spend continue to rise sharply. For instance, in January 2024, it spent $52.9 million on AWS, but that figure ballooned to $176.1 million by December of the same year. Projections estimate that by September 2025, spending on AWS could reach a staggering $518.9 million. This increasingly alarming trajectory emphasizes the unsustainable nature of current business models in the AI sector.
One notable affected entity is Cursor, a coding startup relying heavily on Anthropic’s services. Cursor’s expenses on AWS increased dramatically, jumping from $6.19 million in May 2025 to $12.67 million in June. Amid these rising costs, Cursor was forced to implement price hikes in customer subscriptions, showcasing the ripple effects of Anthropic’s financial struggles.
A Business Model Under Threat
Critics like Zitron firmly assert that Anthropic’s current expenditures are out of control. The concern is that their costs are increasing at nearly the same rate as revenue, which raises alarm bells about the viability of their business model. To achieve profitability, Zitron predicts that a drastic price increase—potentially 100%—might become inevitable. However, such a hike raises the crucial question: is the market ready to absorb such a significant price change?
In summary, Anthropic’s predicament reflects a broader trend across the AI landscape. With ongoing increases in operational costs without corresponding revenue growth, many AI startups face a precarious future. Entrepreneurial ventures in this sector need to reassess their spending habits and funding strategies, as without adjustments, the financial burdens may simply become insurmountable.