Anthropic estimates that customers who violated its terms of service generated 5% to 7% of its revenue during the first nine months of 2026, mainly Chinese AI laboratories using Claude to train competing models. In February 2026, DeepSeek and Moonshot notably used about 24,000 fake accounts to conduct more than 16 million interactions with Claude to collect training data. Anthropic subsequently restricted access and banned tens of thousands of accounts, slowing its growth ahead of a potential IPO, while quarterly revenue reached $16.8 billion in the third quarter. Investor discussions could reduce the proposed valuation from $2 trillion to $1.5 trillion. A draft S-1 also reveals that two clients accounted for 20% of first-half revenue and that the company has substantial long-term commitments for computing resources.
Source: Read the original article
