Anthropic and OpenAI are moving toward the public markets with trillion-dollar expectations. Anthropic investors are modeling an IPO valuation of $2 trillion or more, while OpenAI’s eventual listing could value it at up to $1 trillion.
Their latest private valuation benchmarks are $965 billion and $852 billion.
Open-weight models do not need to beat the strongest proprietary systems to complicate those expectations. If cheaper models absorb enough routine work or force broader price cuts, they can pressure revenue per task and gross margins.
For enterprise buyers, that could also push frontier labs to monetize premium agents and workflows more aggressively.
Cheap models can pressure revenue without winning the frontier
Vercel’s July AI Gateway Production Index found open-weight models handled 29% of gateway tokens in June, up from 11% in April, while accounting for under 4% of spend. Anthropic still captured 61% of spending on 32% of tokens and at least 72% across high-stakes use cases such as coding agents and back-office agents.
Low-cost models are already taking meaningful production volume, but frontier providers still capture most of the dollars. The valuation risk rises if open-weight systems move into higher-value work or customers demand similar price reductions for premium workloads.
OpenAI has already cut prices. On July 30, it cut GPT-5.6 Luna prices by 80% and Terra prices by 20%, describing the reductions as efficiency gains passed on to customers. Lower prices can expand usage, but they also increase the volume and premium revenue needed to preserve growth.
Premium AI has to carry the valuation
Anthropic said its May funding round valued the company at $965 billion and pushed annualized revenue above $47 billion. Its investors are now modeling an October IPO at $2 trillion or more, although senior executives have not set a target.
OpenAI, valued at $852 billion in an employee tender, is now more likely to list next year and could seek up to $1 trillion, the Financial Times reported. Anthropic filed confidential IPO paperwork on June 1, and OpenAI followed on June 8.
The same Financial Times report put OpenAI’s annualized revenue at about $40 billion this month, while Anthropic’s investor materials have projected its first quarterly operating profit. The labs therefore need to defend not just growth, but the margins behind it.
Anthropic’s reported pursuit of a $6 billion Decart acquisition points to one defense: making AI infrastructure more efficient. Products such as Claude Code provide another by tying customer spending to higher-value workflows rather than raw model access.
For investors and enterprise customers, the useful signals are gross margin after inference costs, revenue per completed task, retention, and the share of revenue coming from agentic products. If those metrics hold while open-weight volume rises, frontier labs can absorb cheaper competition.
If premium workloads move down-market or broad price cuts outpace efficiency gains, trillion-dollar valuation assumptions become harder to defend.
Also read: OpenAI and Anthropic are cutting AI costs as open-weight models gain ground, increasing pressure on the economics behind frontier AI.