Anthropic posts $80bn loss in first half of 2026: WSJ

The nonaccomplishment is up $13 cardinal much than past year, driven by accounting charges, the study said.

Anthropic, the artificial intelligence (AI) company known for creating the chatbot Claude, recorded a nett nonaccomplishment of much than $80bn successful the archetypal fractional of 2026, according to the Wall Street Journal, arsenic the marketplace anticipates its looming archetypal nationalist offering (IPO).

The corp besides indicated it could walk up to $750bn successful coming years to get much computing capacity, including imaginable deals presently nether discussion, the Journal said connected Tuesday.

Recommended Stories

• list 1 of 3Why are 8 US states suing OpenAI?
• list 2 of 3AI instrumentality helps stranded protesters find their mode location successful the Indian capital
• list 3 of 3Ex-OpenAI worker accused of planting instrumentality has Saudi backing: Prosecutor

The main operator for the nonaccomplishment was the $72bn successful accounting charges related to changes successful the worth of convertible notes and different fiscal instruments Anthropic issued to rise funding, the study said.

The nonaccomplishment fig was up from $13bn a twelvemonth earlier, according to a draught prospectus, a ineligible papers corporations taxable to fiscal regulators, for its IPO.

The figures amusement the hefty costs of the AI race, adjacent arsenic investors look much intimately astatine whether the wealth companies are pouring into chips and information centres volition wage off.

Anthropic besides shared a database of imaginable hazard factors, including uncertainty implicit however to forestall its products from doing harm, whether done blase cyberattacks oregon blackmail by uncontrolled models.

Anthropic did not respond to a petition for comment, and the study connected the non-public accusation could not beryllium instantly verified.

Investors measurement loss

Reuters archetypal reported successful September that Anthropic posted a nett nonaccomplishment of $42bn successful 2025, arsenic portion of a bid of exclusives from its IPO prospectus.

The institution has benefited from the popularity of Claude Code, its coding adjunct that developers usage to constitute bundle and physique apps.

But immoderate existent and prospective investors who attended Anthropic’s capitalist time successful San Francisco, California, past week are weighing a little valuation for the AI institution aft its latest fiscal disclosures, 4 radical acquainted with the substance told Reuters.

The investors who attended the lawsuit discussed a valuation of astir $1.5 trillion, 3 of the radical said, a important measurement down from the astir $2 trillion valuation antecedently expected for the November stock sale.

Anthropic’s gross jumped to $16.3bn successful the six-month play that ended June 30, up from $1.1bn a twelvemonth ago, portion AI infrastructure spending astir quadrupled to astir $9bn, the Journal reported.

Anthropic reported past week that AI could beryllium months distant from learning to self-replicate connected mundane location devices.