OpenAI and ANthropic, two of the top AI LLM model creators, are deciding on IPOs and are dealing with government restrictions. (Photo by Imen Ben Youssef / Hans Lucas / AFP via Getty Images)
Hans Lucas/AFP via Getty Images
OpenAI and Anthropic spent June proving they are the two labs that will define the next phase of AI, and the month closed with both still moving toward listings near $1 trillion while Washington asserted fresh control over their most powerful models.
Per CNBC, OpenAI is leaning toward delaying its IPO, Anthropic just overtook it on private valuation, and on June 27 the US government decided which of Anthropic's frontier models could return and on what terms. The Big Two are scaling faster than any companies before them, and Washington is now part of the story.
OpenAI signaled this week that it would rather wait. The New York Times reported on June 26 that OpenAI is leaning toward a 2027 listing, easing off a plan to go public as soon as the fourth quarter of 2026, after SpaceX's record IPO cooled and the broader tech market softened.
The company filed a confidential S-1 on May 22 with Goldman Sachs and Morgan Stanley, and Sam Altman has refused to trim the $1 trillion target, a steep climb from the $852 billion mark set when its record $122 billion round closed in March.
Amazon's money sets part of the clock: $35 billion of its commitment unlocks only when OpenAI goes public or reaches AGI, so a long delay carries real cost, and CFO Sarah Friar has told colleagues the company may not be ready, according to the Wall Street Journal.
To widen the base it eventually prices against, OpenAI now sells ChatGPT ads, a business it declared at Cannes Lions, where the US pilot already passed $100 million in annualized revenue on a path the company pegs at $100 billion by 2030 per Reuters.
Anthropic chose this moment to pull ahead on paper.
The company raised at a $965 billion valuation in late May, overtaking OpenAI's private mark for the first time, then filed its own confidential S-1 on June 1. Both filings landed within ten days of each other, a sign the IPO window cracked open by SpaceX is the one every lab is watching.
Anthropic has passed OpenAI in valuation. (Photo by JOEL SAGET / AFP via Getty Images)
AFP via Getty Images
That puts the Big Two on parallel tracks, with Anthropic leaning on enterprise demand and coding tools while OpenAI leans on consumer scale and now advertising, and it ends the long debate over which lab leads with Anthropic in front.
Washington made its presence felt across the Big Two this month. On June 12, the US government issued an export control directive that suspended Anthropic's most capable models, Claude Mythos 5 and Claude Fable 5, for any foreign national, and Anthropic pulled both for every customer to comply.
On June 27, after two weeks of daily talks, Commerce Secretary Howard Lutnick cleared Mythos 5, Anthropic's strongest cybersecurity model, to return to roughly 100 vetted US organizations that defend critical infrastructure, with no export license required.
The split reflects a basic asymmetry: Mythos 5 can be fenced to about 100 named organizations, while Fable 5 reached hundreds of millions of users, which makes any conditional return far harder to police, so Fable 5 stays offline with no timeline. OpenAI faces the same hand: it agreed to roll out its new GPT-5.6 models in phases at the government's request, a launch Altman called bad news, and shared its trusted-partner list with officials first.
The OpenAI and Anthropic story carries direct lessons for any leader who runs operations on frontier models.
Design for model substitution. Washington suspended Claude Mythos 5 and Claude Fable 5, then returned only Mythos 5 two weeks later, which means the most capable model in your stack can move offline on a government timeline. Leaders who route work across OpenAI, Anthropic, and a tested backup keep operating when one vendor goes dark.
Earn a place on the trusted-partner list. Mythos 5 returned for roughly 100 vetted US organizations that defend critical infrastructure, and OpenAI handed officials its trusted-partner list before phasing GPT-5.6. Access to the strongest capability now tracks with vetting status, so the security and compliance posture you build today decides whether you sit near the front of that line.
Read the IPO clock as a vendor signal. OpenAI pushing its listing toward 2027 and Anthropic filing at a $965 billion valuation tell you both labs carry pressure on price, revenue, and disclosure, and that pressure flows into the contracts, support, and product velocity you receive as a customer.
Price in the advertising pivot. OpenAI now sells ChatGPT ads, so any leader who routes brand, customer, or proprietary data through consumer AI tools needs clear rules on what those tools capture and how that data shapes what users see.
Put AI governance on the board agenda. A subpoena from 42 state attorneys general for OpenAI and a model-by-model clearance for Anthropic show oversight moving at the same speed as the labs, which puts your audit trail, vendor due diligence, and usage policy among the decisions executives own now.
The throughline across the Big Two is clear.
OpenAI and Anthropic are pushing revenue, valuation, and capability past every prior benchmark, and government oversight now moves at the same pace, an export order and a clearance granted model by model for Anthropic, a phased model launch and a subpoena from 42 state attorneys general for OpenAI.
Whoever lists first, OpenAI and Anthropic are building the frontier on terms Washington increasingly helps set.
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。