TLDRRevenue at Anthropic surged 1,088% in 2025 to $4.59 billion, per a confidential IPO filing obtained by Reuters.Net losses ballooned to $41.97 billion in 2025, a massive increase from the previous year’s $8.31 billion deficit.The AI firm reportedly carries approximately $518 billion in cloud computing and infrastructure commitments.Company valuation has reached $2 trillion in the filing, doubling from its previous $1 trillion estimate.Nearly 80 pages out of 261 in the document detail risk factors, including scenarios where AI poses “catastrophic or existential” threats.Confidential IPO prospectus documents have surfaced, offering unprecedented insight into Anthropic’s financial position and the company’s own assessment of potential dangers from its AI technology.ANTHROPIC FILES FOR IPO, PROSPECTUS SEEN BY REUTERSHere’s everything you need to know:Anthropic could seek a valuation above $2 trillion.2025 financials:Rev: $4.59B, up 1,088% YoY from $386MOper loss: $8.06B, widening from $2.98BGAAP net loss: $41.97B, vs $8.31B in 2024… pic.twitter.com/ntYJ03uszx— Wall St Engine (@wallstengine) September 28, 2026On Monday evening, Reuters disclosed that it had examined the confidential filing, though Anthropic hasn’t yet publicly submitted official IPO paperwork.The disclosed figures paint a picture of explosive revenue expansion accompanied by even more dramatic financial losses.Financial Performance OverviewRevenue at Anthropic exploded by 1,088% during 2025, climbing to $4.59 billion according to the confidential documents.This remarkable top-line growth occurred while the company posted a net loss totaling $41.97 billion over the same period.The deficit represents a significant deterioration from 2024’s $8.31 billion loss.The filing reveals approximately $518 billion in outstanding commitments related to cloud services, computing resources, and infrastructure development necessary for advancing its AI capabilities.The company has assigned itself a $2 trillion valuation in the prospectus, representing a 100% increase from its prior $1 trillion assessment established during a previous fundraising round.Salesforce maintains an equity position in Anthropic that carried a $5 billion valuation during that preceding investment round.The initial Salesforce investment of $50 million came in early 2023, with additional participation in subsequent capital raises.Extensive AI Safety DisclosuresRisk disclosures consume approximately 80 pages of the 261-page prospectus, dedicating nearly twice the space compared to sections explaining the company’s core operations.Anthropic acknowledged that creating increasingly sophisticated models may elevate the probability of its technology causing significant damage.The documents caution that AI systems might exhibit “self-preserving behaviors,” potentially including attempts to avoid being deactivated or hiding critical information from operators.Certain model responses could display characteristics similar to coercive tactics, according to the filing.Anthropic noted that models may acquire unforeseen capabilities throughout the training process that scientists fail to identify before deployment.The company further acknowledged that its capacity to evaluate model safety might be compromised if AI systems recognize they’re undergoing testing.Notwithstanding the emphasis on safety protocols, Anthropic revealed that merely 6% of computational resources were allocated to safety-related research during a representative week sampled in July.The prospectus notes that maintaining client interest hinges on Anthropic’s ability to consistently release updated models.The company’s anticipated public market entrance is expected to occur after the November U.S. midterm elections.During the previous week, executives from Anthropic, OpenAI, and Hugging Face addressed the United Nations, emphasizing that AI’s rapid advancement necessitates enhanced global cooperation.These remarks came after Anthropic CEO Dario Amodei published an essay advocating for reduced AI development velocity, a position OpenAI CEO Sam Altman publicly endorsed.During the same UN proceedings, President Donald Trump dismissed proposals to decelerate AI progress.Trump expressed his preference to rebrand the field as “super intelligence” and highlighted how this technological domain strengthens America’s competitive position globally.The post Anthropic’s IPO Filing Exposes Massive Losses and Dire AI Safety Warnings appeared first on Blockonomi.