Anthropic's S-1: What a $965B IPO Filing Changes

Anthropic filed a confidential S-1 with the SEC today. The Claude developer submitted the draft registration statement under standard confidential review procedures — the same mechanism Google used before its 2004 debut, and the same law firm (Wilson Sonsini) is advising. The filing follows a $65 billion Series H round that valued Anthropic at $965 billion post-money, leapfrogging OpenAI for the first time.

This is not another funding announcement. It is the formal starting gun for what could become the largest IPO in history — a potential $75 billion raise at a target valuation of $1.75 to $1.8 trillion, with Goldman Sachs, JPMorgan, and Morgan Stanley reportedly in line for lead underwriter roles.

The Numbers Behind the Filing

The financial profile Anthropic brings to Wall Street is unlike anything the public markets have seen from a pre-revenue-to-profit AI company. Run-rate revenue crossed $47 billion in May, and the company has told investors it will exceed $50 billion by end of July — an 80-fold increase in annualized revenue over two years. To put that trajectory in perspective: annualized revenue was $4 billion as recently as July 2025. Expected Q2 2026 revenue of $10.9 billion would more than double the prior quarter.

At $965 billion against that $47 billion run rate, Anthropic trades at roughly 20x sales. For context, that is actually the lowest revenue multiple among the big three AI companies approaching public markets — a detail the bulls cite as evidence the stock is "cheap" at nearly a trillion dollars. The company is on pace for its first profitable quarter, driven by what the Wall Street Journal reported as a projected 130% revenue surge.

The valuation escalation tells its own story. Anthropic closed a $30 billion Series G at $380 billion in February 2026. Four months later, the Series H landed at $965 billion — a 2.5x jump that tracks with product velocity (Claude's successive Opus iterations dominating the Chatbot Arena) and enterprise adoption (Brad Gerstner of Altimeter described "large-scale adoption among the world's most demanding organizations"). In eight months, the valuation moved from $183 billion to $965 billion. That is not a growth curve — it is a vertical line.

The Series H investor list reads like a sovereign wealth fund conference: Altimeter Capital, Dragoneer, Greenoaks, Sequoia, Capital Group, Coatue, D1 Capital Partners, Baillie Gifford, Blackstone, Brookfield, D.E. Shaw, DST Global, and Fidelity. Strategic infrastructure partners Samsung, SK Hynix, and Micron also participated — a signal that the chip supply chain is betting on Anthropic's compute demand lasting well beyond the IPO window.

Why Confidential, Why Now

A confidential S-1 lets Anthropic begin the SEC review process without immediately disclosing revenue, margins, cost structure, or risk factors to competitors. The filing becomes public only 15 days before the roadshow — giving the company months to negotiate terms and iterate on disclosures while OpenAI watches from behind.

The timing is strategic. Anthropic is positioning itself ahead of OpenAI in what multiple outlets call the most dramatic AI IPO wave in history. OpenAI was reportedly preparing its own confidential filing targeting a fall 2026 debut. SpaceX is in the same pipeline. Together, these three companies could introduce more than $3 trillion in market capitalization into public markets in a single season.

The "IPO before the market sneezes" reading — a phrase from the HN thread that hit 463 points and 377 comments — captures the real calculus. The venture capital math is straightforward: in Q1 2026 alone, global VC firms invested roughly $300 billion into about 6,000 startups, with 80% of that capital flowing into AI. That capital needs an exit. If you are sitting on $965 billion in private valuation and the public window is open, you file. The alternative — waiting for rate hikes, geopolitical disruption, or a model-capability plateau — is a risk that no fiduciary can justify when the paperwork is ready.

The Prediction Markets Have Already Priced This

Anthropic's dominance is no longer a matter of debate in the markets that track it. On Polymarket, Anthropic holds an 83% probability of having the best AI model by end of June, with Google at 13.5% and OpenAI at 3.3%. The $500B+ valuation in 2026 market sits at 98% — a near-certainty that was already priced in before the Series H.

This is the surface that should concern serious investors. When every prediction market, every lead investor, and every product metric points in the same direction, the asymmetric risk is no longer to the upside. As we covered in our analysis of Anthropic at 92% across prediction markets, the consensus is so complete that the only question left is what breaks it.

The Contrarian Flags

Three signals sit beneath the bullish consensus:

The Burry Warning. Investor Michael Burry said on social media there is "no guarantee" Anthropic gets close to a $1 trillion price tag. His argument: building frontier AI models is "far too expensive," and compute could end up more like a commodity. This is the same structural bear case that haunts every capex-intensive technology cycle — the question is whether inference margins hold or compress.

The Risk Transfer. Bank of America's assessment of the triple-IPO wave is blunt: this cycle is "essentially a large-scale transfer of accumulated risk from early investors to the public market." In the first quarter of 2026, global VC firms invested roughly $300 billion into ~6,000 startups, with 80% flowing into AI. That capital now needs an exit. The S-1 is the exit door.

The Pentagon Problem. Anthropic is locked in a legal battle after the Pentagon declared it a supply-chain risk — a designation typically reserved for foreign adversaries. The company has said the Trump administration's move could jeopardize billions of dollars in revenue. For a company that has positioned safety as its core differentiator — Dario Amodei's Responsible Scaling Policy is not marketing, it is an operational constraint — a government adversarial posture creates a paradox. The same safety commitments that justify the premium valuation are apparently the same commitments that put Anthropic at odds with the national security establishment. This is the kind of risk factor that a public S-1 will have to disclose in detail, and it is the one that current private-market valuations may not fully reflect.

The Index Inclusion Trap

The HN discussion surfaced a structural concern that retail investors should understand. Recent changes to NASDAQ and CRSP index rules have shortened the path from IPO to index inclusion from months to as little as 5-15 days. Float requirements dropped from 10% to roughly $3.3 billion market cap. Multi-class share structures now count for eligibility.

The implication: if Anthropic IPOs at $1.75 trillion with sufficient float, it could enter the S&P 500 or major NASDAQ indices within weeks. Index funds and 401(k) plans would be forced to buy — not because portfolio managers chose to, but because the index rules demand it. As one HN commenter put it: the plan is to "rapidly drive prices up, get companies listed in NASDAQ so funds are forced to purchase them at higher prices, then leave retirement accounts holding the bag."

This is not conspiracy. It is mechanics. And it is the mechanism by which a $965 billion private valuation becomes a $1.75 trillion public valuation without requiring a single new institutional conviction.

What the S-1 Will Actually Reveal

When the filing goes public — likely 15 days before the roadshow, targeting October 2026 — it will answer questions that the private market has been able to avoid:

Margin structure. Anthropic's $1.25 billion monthly GPU commitment to SpaceX through May 2029 is a known obligation. What is not known is the inference gross margin, the training-to-inference cost ratio, or how sensitive profitability is to model generation turnover. The bull case requires margins to expand; the bear case argues that compute is a commodity and margins compress.

Customer concentration. How much of $47 billion in run-rate revenue comes from Amazon and AWS-dependent workloads? The $100 billion AWS commitment is both Anthropic's greatest asset and its greatest dependency. Amazon's $5 billion contribution to the Series H — part of the $15 billion hyperscaler tranche — deepens this entanglement. The S-1 will have to quantify the concentration, and the market will have to decide whether it is a moat or a single point of failure.

The safety cost. Anthropic has positioned itself as the safety-first AI lab. Dario Amodei's Responsible Scaling Policy is genuine differentiation — but it is also a cost center. The S-1 will quantify, for the first time, what safety research costs as a percentage of revenue, and whether the market is willing to pay for it.

Government risk. The Pentagon supply-chain designation will require a detailed risk disclosure. The legal costs, revenue at risk, and strategic implications will be public record.

The Google Comparison — and Why It Might Not Hold

In the HN thread, one commenter drew the obvious parallel: Anthropic resembles Google's 2004 IPO — strong growth, real margins, a dominant product, and doom narratives that turned out to be noise. The comparison is tempting. Google went public at $23 billion and was worth $2 trillion within two decades. The "this is overpriced" chorus was loud and wrong.

But the counter-argument is equally instructive. The survivors of technology transitions — Amazon, Google, Apple — emerged from brutal selection cycles where the vast majority of well-funded competitors did not survive. As one HN commenter noted, "the real survivors emerge after busts," suggesting that betting on the current frontrunner is less reliable than betting on "successors." Whether Anthropic is the Google of AI or the AltaVista is a question that cannot be answered by revenue multiples alone. It depends on whether the moat is in the model, the data, the distribution, or the team — and which of those the market is actually pricing.

What This Changes

Anthropic's S-1 marks the transition from a private AI capital cycle — where valuations are set by sophisticated investors who can wait years for returns — to a public one, where quarterly earnings, margin guidance, and analyst downgrades drive the conversation. The valuation escalation from $183 billion to $965 billion in eight months happened in a world without public scrutiny of unit economics. That world ends when the S-1 becomes public.

For operators building on Claude, the IPO changes the incentive structure. A public Anthropic is accountable to shareholders, not just to the mission statement. Pricing pressure, margin targets, and growth expectations will shape API economics in ways that a private company could defer. The question is whether the same company that priced Claude competitively to win developer share will maintain that stance when Wall Street is watching gross margins. The Polymarket pricing of Anthropic at 83% for best model reflects product leadership — but product leadership and margin-friendly pricing do not always coexist in public companies.

For the broader AI market, the three-way IPO wave is a stress test. If Anthropic, OpenAI, and SpaceX collectively add $3 trillion in public market value during a period of already-elevated tech valuations, the concentration risk in tech equities reaches a level the market has never tested. As the Yahoo Finance analysis noted, these IPOs will "become a powerful referendum on the sustainability of the current technology boom."

If the S-1 numbers hold up under analyst scrutiny, October could mark the beginning of a new era in public-market AI investment. If they don't, the retelling will focus on a different phrase from that HN thread: "mad rush to IPO before the market sneezes."

The S-1 is filed. The clock is running. What happens next is no longer up to Dario Amodei — it is up to the market.