Our Take
Government restriction on a model validates rather than punishes the company—a rare outcome that works because Anthropic's publicly available models are already winning market share before the ban took effect.
Why it matters
Enterprise buyers are choosing Anthropic over OpenAI for the first time, driven by both product performance and a contrarian reputation. The timing matters because Anthropic is preparing an IPO; the ban creates a narrative asset instead of a liability.
Do this week
Procurement: audit your AI vendor lock-in before the IPO filing resolves—Claude Opus adoption may shift pricing or support terms post-public.
Anthropic's market share surpasses OpenAI in business spending
Anthropic claimed more than 41% of AI subscriptions purchased by businesses in May, up 2.5 percentage points from April, according to spending data from Ramp covering over 70,000 companies (per Ramp analysis). OpenAI held 39.5%, essentially flat month-over-month. The majority of business spending flows through API calls—token use for coding and other tasks—where Anthropic's Claude Opus models dominate the visible transaction data.
Most of this adoption predates the White House ban. On Friday, the Trump administration ordered Anthropic to block non-Americans, including its own employees, from accessing Mythos 5 and Fable 5, its latest high-capability models. Mythos had been in limited release since April; Fable 5 launched three days before the ban and was pulled immediately. The administration cited export control directives, though reporting suggests the trigger was Fable 5's weak guardrails against accessing Mythos capabilities, a model Anthropic itself marketed as too dangerous for public release due to its code-vulnerability detection strength.
Anthropic remains publicly available on Opus 4.8, released in late May, which accounts for the bulk of enterprise spend.
The ban may amplify, not weaken, Anthropic's appeal
Ramp's lead economist, Ara Kharazian, told TechCrunch the government action could help Anthropic. "Anthropic's best month on record, as far as business adoption, was the month that the Department of Defense labeled them a supply-chain risk," in March (company-reported context from source). The pattern repeats: friction with the state creates a halo of capability and principle.
This works because the models customers actually use—Claude Opus variants—remain unaffected by the ban. Mythos and Fable 5 were not yet embedded in enterprise workflows. The withdrawal removes a potential liability (public models with weak controls) while the government's own framing ("too dangerous") reinforces Anthropic's positioning as the model that can do hard technical work. Buyers see certification of capability at no cost to their own deployments.
Anthropic filed confidential IPO paperwork in early June after raising $65 billion at a $965 billion valuation in late May. Public-market investors typically flee companies in regulatory conflict. Here, the conflict may instead signal moat and scarcity—the opposite of a burden.
Lock in your Claude terms while adoption is rising
If you use Claude API, your consumption cost and support tier are about to matter more. Anthropic's business momentum is real and measurable; the IPO filing and government drama will create pricing pressure and contract pressure in the next 12 months. Negotiate multi-year API pricing and support SLAs before the public offering closes and demand rises faster than supply.
Opus 4.8 remains your reliable production choice. Do not wait for Mythos or Fable to stabilize; your cost and availability risk is lowest if you commit to Opus now.