Anchor story: On Monday, October 26, 2026, Anthropic filed its Form S-1 registration statement with the SEC, setting up a Nasdaq listing under the ticker ANT and — per reporting — a possible raise of $100B or more at a $1.5–2T valuation, none of which is yet in the public filing.
Why it matters now: The filing is the clearest dated evidence that the AI build-out is still accelerating: $16.3B of H1-2026 revenue, ~$417B of committed compute spend, and a $15B Amazon financing pipeline — a company pricing itself against a compute bill that now dominates its cost base.
Practical lesson: You can't underwrite the company before it prices. But you can deliberately decide whether the build-out theme — power, chips, data centers, and the debt behind them — is in your portfolio.
Charon: Welcome to Ad Valorem Syndicate Room, episode 53. I'm Charon. Today, the biggest AI lab in the private world walks up to the public window.
Kore: Thanks, Charon. On Monday, October 26, Anthropic filed its S-1 registration statement with the SEC, setting up what could be one of the largest stock debuts in history.
Charon: Let's set the stakes, because the numbers here are not normal. This is a five-year-old company. In the first half of 2026, it generated $16.3B of revenue, up from $1.2B a year earlier. That is 14x growth in a single year.
Kore: And the growth rate itself is the surprise. Revenue in the quarter ended June 30 rose 1,368%, year over year. In the second quarter alone, it spent $5.6B on compute and generated $11.6B in revenue — about $2 of revenue for every dollar of compute cost.
Charon: Now, the cost side, where the risk lives. The same filing shows a net loss of $80.7B for that half, versus about $13B a year earlier. But here's the nuance the headlines miss. Roughly $72B of that is a non-operating accounting charge tied to convertible bonds at a soaring valuation. The operating loss actually narrowed to about $693M, and adjusted operating income swung positive to $1.6B.
Kore: That distinction matters for anyone allocating. One is a paper charge driven by the company's own valuation. Two is the business not yet covering its cash costs. The filing is explicit that losses will continue.
Charon: And why is it raising this much? The prospectus details future commitments to pay compute providers approximately $417B — data centers, the chips inside them, and cloud services — with deals running up to 20 years. This company has pre-committed most of its next decade of capex before it has priced a single public share.
Kore: There is also fresh liquidity behind it. Amazon has agreed to provide up to $15B in financing, available in $5B tranches. So the balance sheet story is not just the S-1 — it is a committed pipeline.
Charon: The stock will list on the Nasdaq Global Select Market under the ticker ANT. Goldman Sachs has the lead-left role, with Morgan Stanley co-leading. The offering is expected to target the week of November 16, and reports point to a possible raise of $100B or more at a valuation between $1.5T and $2T. None of those numbers are in the public filing yet — they will be set after the roadshow. Treat them as reported expectations, not terms.
Kore: One structural detail deserves its own section, because it departs from every textbook IPO. The lock-up is staggered. Insiders can begin selling at day 27, day 48, day 62, and day 76, instead of the traditional 180-day cliff. That is a real liquidity design choice, and it will shape post-pricing volatility.
Charon: Let's connect this to the build-out we've been tracking all season. At Oracle AI World on October 28, Oracle added 850 MW of AI capacity and reported OCI revenue of $7.4B, up 121% year over year. Anthropic is pricing itself into a world where the compute bill is the dominant line item. When $2 of revenue per compute dollar is the headline, the margin story is really a power story and a chip story.
Kore: The same week, the rails for moving value were scaling too. Solana reported 14 million stablecoin holders and 1.5 million addresses holding tokenized equities, and Securitize extended its institutional tokenization to Stellar on October 28. The infrastructure that underwrites AI is being financed, and the rails that settle it are maturing, on the same timeline.
Charon: So what's the lesson for a smaller investor? Don't underwrite this company. You can't buy it pre-listing, and when you can, the valuation will be set by institutions. But you can underwrite the theme. Every dollar of that $417B compute commitment lands somewhere — data centers, power, chips, cooling, networking, and the debt that finances all of it. The filing is the clearest, most dated proof yet that the build-out is still accelerating, and that the bottleneck has moved from models to megawatts.
Kore: The downside to hold against that thesis is simple. If growth decelerates, the pre-committed capex becomes a fixed cost on a slower top line, and the staggered lock-up means insiders have options that early public holders do not. That is the risk in one sentence.
Charon: Our suggestion — and this is not advice — is to keep one concrete step in your week: read the S-1's risk factors, specifically the pages on safety and commercial tradeoffs, and map which segments of the compute stack you actually own. Then decide whether the build-out thesis is in your portfolio, deliberately. Position before you predict.
Kore: For sources, the primary document is the SEC filing itself, dated October 26, with coverage from Reuters, CNBC, and Bloomberg. The full reference list is in the episode notes. This content is AI-generated, for informational purposes only.