According to reporting from the Wall Street Journal, relayed by Reuters, Anthropic recorded a net loss exceeding $80 billion during the first half of 2026. If confirmed, this figure would rank among the largest losses ever reported by a technology company over such a short period, far surpassing the losses typically disclosed by generative AI startups to date.
The scale of this loss should be understood in the context of the massive capital outlays frontier AI labs are committing to train and operate their models. Between raw compute costs, multi-year contracts with chip and cloud providers, and forward commitments made to secure future server capacity, the charges booked by these companies can include elements not strictly tied to day-to-day operations, such as provisions or prepaid contractual obligations valued on an accrual basis.
Anthropic, the maker of the Claude model family, has in recent months pursued a series of large funding rounds and infrastructure partnerships as part of an industry-wide race for compute capacity among leading labs, OpenAI chief among them. Such spending is usually paired with rapid growth in subscription and API revenue, but the gap between revenue and costs remains substantial across most frontier labs at this stage.
It is worth noting that the precise accounting methodology behind this figure has not been made public in the information available so far. Numbers of this magnitude, when reported by financial press based on internal documents or sources close to the matter, warrant some caution until Anthropic or its investors confirm the data officially. The report nonetheless feeds into an ongoing debate about the financial sustainability of leading generative AI labs' business models, at a time when their valuations keep climbing despite substantial operating losses.