Anthropic Files for an IPO: What the S-1 Shows, and What It Means for SPY Traders

On October 26, 2026, Anthropic, the maker of the Claude AI models, publicly filed its registration statement on Form S-1 with the U.S. Securities and Exchange Commission. It is one of the most anticipated listings in years. This is an evaluation of what the filing shows, what it leaves open, and what it means for traders, written for education and not as a recommendation to buy or sell anything.

The filing in brief

Until the price range is set in an amended filing, nobody can compute a valuation from the prospectus itself. Every valuation number you see before then is a report or an estimate.

Revenue: the growth story

• First half of 2026: $16.3 billion of revenue, which the filing puts at about 1,300% growth from the same period a year earlier.
• Full year 2025: about $4.6 billion, as reported in coverage of the filing.
• Simple run rate: doubling the first half gives about $33 billion a year, before any further growth.

Growth like this at this size is rare. The questions an analyst asks next are how concentrated that revenue is, how much of it comes through cloud partners, and what it costs to deliver.

The $80 billion loss, and why the headline misleads

The filing reports a net loss of about $80.7 billion for the first half of 2026. Most of it, about $72.4 billion, is a change in the fair value of convertible notes, derivative liabilities and SAFEs. When the company's estimated value rises, the accounting value of those instruments rises too, and the increase is booked as a loss. It is a non-cash charge, and it typically stops once those instruments convert into stock at the listing.

The more useful numbers sit above that line. Press coverage of the filing reports a GAAP operating loss of roughly $0.7 billion for the half, sharply smaller than a year earlier, and about $3.9 billion of cash generated by operations. The company also reports about $112 billion of cash and cash equivalents and $19 billion of short-term investments at June 30, 2026.

The compute bill

Frontier AI models are expensive to train and to serve. Coverage of the filing highlights very large multi-year compute and equipment commitments, including a reported agreement to buy GPU capacity from SpaceX worth about $3.1 billion a month once fully delivered. Commitments of this size are the main reason cash from operations today says little about free cash flow in the years ahead. Whether revenue keeps outrunning compute costs is the central question for the stock.

Governance: what public shareholders don't get

Anthropic is a Delaware public benefit corporation, which lets its board weigh its stated mission and the people its products affect alongside shareholder returns. Two features stand out:

• A founders' share. Reports on the filing describe a single Class F share, held by an entity of the seven co-founders, carrying 50.1% of the vote on most shareholder matters. That makes Anthropic a controlled company under Nasdaq rules.
• The Long-Term Benefit Trust. A trust with no economic stake elects the directors who will hold a majority of the board.

Buyers of Class A shares get one vote per share, but the filing itself warns that they generally cannot change the outcome of a vote. Investors who accept that are betting on the business and its leadership, not on their own ability to influence it.

The valuation debate

Press reports have cited a possible valuation of roughly $1.5 to $2 trillion, up from a reported $965 billion post-money private round in May 2026. On the simple annualized first-half revenue of about $33 billion, that range is about 45 to 60 times sales. The bull case is that revenue is still growing several times over each year and the operating loss is close to breakeven. The bear case is that the multiple already prices in years of that growth, against compute commitments that grow with it and competitors with deep pockets. Both cases are reasonable; the price range, when it is set, will show which one the underwriters think the market will pay for.

What it means for SPY and options traders

• No direct effect on SPY at first. SPY holds only S&P 500 members, and S&P 500 eligibility includes a record of positive GAAP earnings. A newly listed company with a large net loss is not a near-term candidate.
• Indirect effects are possible. A listing this size draws money from other large technology stocks as funds make room, and it becomes a sentiment gauge for the whole AI trade, which weighs heavily in the S&P 500.
• Listing day volatility. Expect a wide opening range, thin early liquidity and sharp reversals. Market orders on the first day are a common and expensive mistake.
• Options come later. Listed options usually start a few trading days after the first trade, typically with very high implied volatility and wide spreads. Selling premium on a stock with no trading history means sizing for moves nobody has measured yet.
• Mark the calendar. The pricing, the first trading day and, months later, the end of the lock-up period, when insiders may sell, are dates when broad market volatility can pick up.

Our take

The filing shows a company with extraordinary revenue growth, a near-breakeven operating line, a very large cash balance, and governance that keeps public shareholders out of control by design. The headline loss is mostly accounting. The real risks are the cost of compute, competition, and a valuation that leaves little room for disappointment. For SPY traders the practical point is simpler: treat the pricing and the first trading days as event days, when the market can move more than usual.

Figures come from the S-1 and same-day press coverage of it. Where a number appears only in press reports, the article says so. The terms may change in amended filings before the offering. Anthropic develops the AI models that we, like many traders, use to build software; we hold no position in it and receive nothing from it.