Anthropic confirmed this morning that it has confidentially submitted a draft Form S-1 to the U.S. Securities and Exchange Commission for a proposed initial public offering. The announcement came from Anthropic itself, not a leak, not a rumor, and was confirmed within hours by Reuters, CNBC, NPR, and US News. The valuation in play is roughly $965 billion based on the $65 billion round that closed late May. That makes it one of the largest IPO setups in history and the first frontier AI lab to formally start the public-market clock. The bigger story underneath the filing is who is no longer in the lead. Prediction markets had OpenAI going public first, with most participants betting that Sam Altman's company would set the template that Anthropic followed. They were wrong. Anthropic just edged ahead of OpenAI in the most closely watched race in the AI industry, and it did it the same week it released its most capable model yet, signed a $47 billion run-rate, and was named alongside the Pope on a global ethics platform. The momentum is now compounding visibly. Today is the day the public market gets its first real shot at owning Claude.

Here is what we know on the record, with sources.

The filing is real and confidential. Anthropic submitted a draft Form S-1 with the SEC under standard confidential review procedures. Confidential filing means the SEC reviews the document privately before any public release. Anthropic described it as giving the company "the option to go public" once review is complete. Share count and price are not yet set. Trading is months away.

$965 billion valuation. This is the post-money number from the $65 billion Series H round that closed late May. Up from $380 billion in February. Up from $61 billion last year. The valuation roughly doubled in three months, then doubled again over the following two. That trajectory is the part underwriting today's filing.

$47 billion annualized revenue run-rate. Reported by Reuters and others today, up from $14 billion in February. A run-rate of this size, growing this fast, is what makes Wall Street confident the company can actually trade at a number near $1 trillion without immediately losing 40% of its market cap. The revenue justifies the price in a way that almost no historical IPO at this valuation has been able to claim.

Beat OpenAI to the filing. Reuters' lede was "edging ahead of rival OpenAI in a closely watched race to reach public markets." On prediction markets earlier this year, most participants had OpenAI going first. OpenAI is still expected to file confidentially within weeks for a September listing at $852 billion to $1 trillion. SpaceX filed publicly May 20 and starts trading June 12 under SPCX at up to a $1.75 trillion valuation. The trifecta is now confirmed.

Compute commitments remain enormous. As part of the SpaceX deal disclosed in SpaceX's own prospectus, Anthropic is paying SpaceX $1.25 billion per month through May 2029 for Colossus 1 compute. That single contract alone exceeds $50 billion over its life. The numbers behind today's filing tell investors Anthropic believes its revenue will compound through that compute bill and out the other side.

No financial disclosure yet. The confidential filing means Anthropic does not yet have to publish revenue, margins, customer concentration, or cost structure. That happens later when the S-1 goes public ahead of the actual offering. Today is the start of the formal SEC clock, not the moment the books open.

Because the position they have built over the last six months was always going to make them the first to file when the market opened. Look at what happened since February:

In February, Anthropic raised $30 billion at $380 billion. In late May, they closed $65 billion at $965 billion. They went from $14 billion run-rate to $47 billion in the same window. They announced a SpaceX compute deal that locked in years of capacity. They poached Andrej Karpathy from OpenAI. They had their co-founder stand next to the Pope. They shipped Claude Opus 4.7, then Opus 4.8 in 41 days, the fastest cadence in their history. They committed to bringing Mythos-class models to all customers "in the coming weeks." They told investors they were on track for their first quarterly operating profit.

That is the resume Wall Street wanted to see before a frontier-AI IPO. OpenAI is still working through structural questions, the Musk lawsuit aftermath, the for-profit transition, ChatGPT's commercial model, the consumer voice product running on a years-old model. Anthropic is cleaner. Enterprise-first, fewer governance complications, faster product cadence, and a customer base of Fortune 500 buyers who actually pay seven and eight figures. They were always more IPO-ready. Today they cashed it in.

First, this is the moment AI valuations stop being private hand-shakes and start being public-market discoveries. A $965 billion private valuation, set by Sequoia and Dragoneer with limited transparency, becomes a real price once the SEC clock starts. Underwriters will probe the cap table, the revenue concentration, the compute spend, the unit economics, the customer attrition. The numbers Anthropic told investors privately are going to be tested by Wall Street. Whether the $965 billion holds, gets repriced higher, or comes down at IPO will tell us what the actual market value of a frontier AI lab is. That answer is the most important number in tech right now.

Second, this is the moment retail can finally participate. For two years public investors have been forced to express an AI thesis through Nvidia, Microsoft, or Google. Anthropic's IPO will be the first chance to own a pure-play frontier AI lab in liquid public markets. Cerebras was an AI chipmaker. SpaceX is a rocket company that owns an AI lab. Anthropic, on the other hand, is the AI company. The model maker. That distinction is going to matter enormously to how the listing prices and trades.

Third, the IPO trifecta is now fully visible. SpaceX trades June 12. Anthropic just filed and could list in October. OpenAI is expected to file within weeks for a September target. Within four months, three of the biggest names in AI will be public companies, or at least within sight of trading. The era of "AI lives in private markets" ends inside the next two quarters. That changes how every other AI company gets valued, how every employee equity package gets priced, and how every enterprise buyer thinks about vendor risk.

The honest forward map.

The SEC will review the confidential filing privately for several weeks. Anthropic can amend it during that window without public disclosure. When the company is ready to actually go public, the S-1 becomes public, the roadshow begins, and pricing typically follows within a few weeks of that. Realistic timing for an actual listing is fall 2026, most likely October based on Reuters' reporting.

Watch three things in the meantime. One, whether OpenAI's expected filing comes in time to launch first in September, leapfrogging Anthropic's October target. Two, how SpaceX's June 12 debut performs, because it sets the appetite gauge for the rest of the trifecta. Three, whether the public S-1, when it lands, shows the cost structure investors are hoping for, or whether the SpaceX compute bill and the talent costs reveal margins thinner than private valuations have assumed.

For now, the takeaway is simpler. The most valuable private AI company on earth just filed its IPO before the company everyone thought would go first. That is the line.

Quick personal note from us.

Today is the perfect day to launch this, because the entire reason Anthropic just filed at $965 billion is that frontier coding models have become the most valuable category in AI. The question that matters for anyone who actually builds is which one is best for your code, not which one tops a generic leaderboard.

That is what we built World of AI Bench to answer.

It is a vibe coding benchmark that compares frontier AI models on real coding tasks. Live previews of the output. Private benchmarks you can run on your own prompts. Cost tracking so you can see what each model actually costs to finish a job, not just per-token pricing. Community rankings so you see what other builders are choosing. Everything in one place, designed for the way people actually code with AI in 2026.

We are live on Product Hunt today. If this newsletter has been useful to you over the last few months, a vote would mean a lot. It is the single thing that gets the tool in front of more builders today.

👉 Vote on Product Hunt: https://www.producthunt.com/products/world-of-ai-bench?launch=world-of-ai-bench

👉 Try the bench: https://www.woaibench.ai/

Either way, thank you for reading. Now back to the recap.

Anthropic confidentially submitted its draft Form S-1 to the SEC today, June 1, 2026, in a planned IPO at a roughly $965 billion valuation, the largest frontier AI listing setup in history. The company is on a $47 billion annualized revenue run-rate, up from $14 billion in February, and just raised $65 billion in May. It beat OpenAI to the filing in defiance of prediction markets that had Sam Altman's company going first. SpaceX starts trading June 12, OpenAI is expected to file within weeks for a September target, and the AI IPO trifecta is now fully visible inside the next four months. For the public market, this is the first time you can own a pure-play frontier AI lab in liquid stock. For the industry, this is the moment private valuations get tested by Wall Street. The race is no longer who builds the best model. The race is who gets to set the price of AI for the next decade. Today Anthropic stepped to the front of that line.

World of AI Benchmark 👉 Here

Anthropic’s S-1 Notice 👉 Here

Stay building. 🤖