Today — Wednesday, 14 October 2026 — Anthropic is presenting its third-quarter financials to institutional investors at its San Francisco headquarters. CNBC reported the plan on 13 October, citing two people familiar with it; Bloomberg first reported the investor day on 1 October. It is the first time the Q3 numbers will be shown to prospective investors, ahead of a possible mid-November IPO that could value the company near $2 trillion. Anthropic declined to comment, so treat everything below as reported — not confirmed by the company.
For anyone watching AI stocks, this is a rare look at the engine room of the world’s most-watched private AI company. But AI-lab financials are a hall of mirrors: revenue figures come in several flavours, losses carry huge accounting adjustments, and forward-looking estimates get quoted as if they were earned money. Here is how to read the Anthropic story like an investor — before the hype reads you.
What is happening today?
Anthropic is meeting a select group of institutional investors at its SF headquarters, showing Q3 financials ahead of a possible IPO marketing push starting the week of 9 November. The reported timeline, from Bloomberg’s sources and CNBC’s 13 October report: formal marketing could begin the week of 9 November, with shares trading before Thanksgiving on 26 November, though the timetable remains provisional. The company confidentially filed its S-1 with the US SEC earlier in 2026, but the financials are not yet public.
The scale being discussed is historic. Prospective investors put fair value at $1.8 trillion to $2 trillion, per Bloomberg; Reuters’ review of the prospectus suggested more than $2 trillion. The May 2026 private round valued Anthropic at $965 billion — so a listing at $2 trillion would more than double that in months. The raise could reach roughly $100 billion, which would make it one of the biggest IPOs ever, ahead of SpaceX’s record $1.77 trillion debut in June/July 2026. Morgan Stanley, Goldman Sachs, JPMorgan and Citigroup are leading the deal.
What do we actually know about Anthropic’s numbers?
Only reported figures exist — the S-1 is still private, so every number below carries its source and a caveat.
| Figure | Source and date | Caveat |
|---|---|---|
| 2025 revenue: ~$4.6 billion (12x the prior year) | Reuters prospectus review, late Sep 2026 | Reported, not confirmed by Anthropic |
| 2025 operating loss: ~$8.06 billion | Same prospectus review | Reported; excludes the accounting adjustments below |
| 2025 net loss: ~$42 billion | Same prospectus review | Over $34 billion of it is non-cash fair-value liability adjustments — accounting, not cash burn |
| Computing spend 2025: $7.3 billion = 58% of $12.65 billion total operating expenses | Motley Fool analysis of the prospectus | Shows where the money really goes |
| Annualised revenue run rate: $65 billion (July 2026); expected over $100 billion by year-end | People familiar with Anthropic’s plans, via CNBC, Oct 2026 | Forward-looking estimate — a projection, not earned revenue |
| Q2 2026: positive on an adjusted operating income basis | Secondary reporting | A claimed milestone — first frontier AI lab to report this — but on Anthropic’s chosen definition |
| $518 billion of multi-year cloud and data-centre commitments, ~80% non-cancelable, over $100 billion tied to AWS | Reuters prospectus review | This is the number that makes or breaks the valuation story |
The OpenAI lesson: why the same company can quote $50B and $70B
Revenue definitions are negotiable — and the same company’s top line can change by tens of billions depending on what gets counted. OpenAI recently told current investors its annualised revenue was $50 billion — below the $70 billion widely reported earlier. The updated figure excluded revenue from cloud partners; a person familiar with the matter told CNBC the change did not reflect weaker demand. The takeaway for today: when Anthropic’s Q3 figures eventually leak or get filed, ask whether the number includes or strips out cloud-partner revenue, API credits, and partner deals. OpenAI’s example shows how two honest-sounding figures for one company can differ by $20 billion purely on accounting choices.
4 checks before you trust any AI-lab number
Run every headline figure through these four filters and most of the hype falls off.
1. Earned revenue or annualised run rate?
A “$65 billion run rate” multiplies one good month’s annualised revenue by twelve. It is a real number with real momentum behind it, but it is not $65 billion sitting in the bank — and the “$100 billion-plus by year-end” figure quoted by people familiar with Anthropic’s plans is a forecast, not a result. Treat run rates as direction, not destiny.
2. What is excluded from the revenue number?
OpenAI’s $50B-versus-$70B gap came entirely from stripping cloud-partner revenue. Anthropic has over $100 billion tied to AWS in its commitment stack. Whenever a Q3 revenue figure surfaces, the first question is the accounting boundary: which partner revenues and credits are in, and which are out?
3. Operating loss or net loss?
Anthropic’s 2025 net loss of $42 billion sounds catastrophic until you learn that more than $34 billion of it is non-cash fair-value adjustments on liabilities — a revaluation tied to past fundraising, not money leaving the company. The operating loss of $8.06 billion is the closer measure of the actual cost of running the business. Always separate accounting charges from cash burn.
4. What sits underneath the growth?
The $518 billion of infrastructure commitments — roughly 80% of it non-cancelable — is the number that decides whether a $2 trillion valuation is reasonable or reckless. Non-cancelable means Anthropic owes this money whether demand holds or not. It also explains the capital-market machinery around the IPO: Broadcom has begun assembling $60 billion in debt to help finance AI chips for Anthropic and others. Broadcom CEO Hock Tan warned on 12 October that rising interest rates could make access to capital the biggest constraint on AI companies — and chip stocks fell the next day. When a lab’s commitments are measured in hundreds of billions, the cost of borrowing is part of the business model.
Anthropic vs OpenAI: what has been disclosed
The two frontier labs are now disclosing in parallel — and the contrast is the story.
| Anthropic | OpenAI | |
|---|---|---|
| IPO timing | Formal marketing possibly from the week of 9 Nov; trading before Thanksgiving (reported, provisional) | IPO postponed; raising at least $30B at ~$1.4T (Bloomberg) |
| Revenue story | $4.6B in 2025 → $65B run rate July 2026 (reported) | $50B annualised, excluding cloud-partner revenue (company-disclosed) |
| Losses | ~$8B operating / ~$42B net in 2025 (reported, mostly non-cash) | Not disclosed in this round |
| Valuation chatter | $1.8–$2T (reported) | ~$1.4T (reported) |
| Commitments | $518B multi-year, ~80% non-cancelable (reported) | Not detailed in this round |
Every figure in this table is as reported by the cited outlets — neither company has confirmed these numbers publicly, and Anthropic’s S-1 financials remain private.
What this means if you want a piece of it
The honest version: ordinary investors cannot buy before the listing — and the discipline that protects you on listing day is decided before listing day. This is an investor-literacy piece, not financial advice; it cannot tell you what to buy.
- You cannot get in today. The 14 October meeting is for institutional investors only. Retail access starts on or after the first day of trading, through any brokerage with US market access.
- Decide your price discipline now. Record-breaking IPOs routinely spike on opening day and then settle as the first public filings get scrutinised. Knowing what you would pay — before the ticker moves — is the only edge retail investors get.
- Read the S-1 when it goes public. The prospectus will show revenue recognition, customer concentration, compute commitments, and the founder-control structure (the filing creates a Founder LLC/Class F voting structure concentrating control with the founders). That document is the real earnings call.
- Watch the financing weather, not just the company. Hock Tan’s 12 October warning and the Oct 13 chip-stock slide matter here: a $2 trillion listing priced for perfect conditions is vulnerable to rising rates. Anthropic’s timing — before Thanksgiving, before December’s market slowdown — is itself a bet on the window staying open.
Want more context on Anthropic’s products before you watch its finances? Our Claude Haiku 5.5 review breaks down what its cheapest model actually buys you, and our AI agent safety checks guide walks through the incident classes investors are quietly pricing in. For how AI plugs into your own tools, start with MCP explained.
Frequently asked questions
Can I buy Anthropic stock before the IPO?
No. The 14 October investor day and any pre-listing marketing are for institutional investors. Retail investors get access on or after the first trading day through a brokerage with US market access.
Is the $2 trillion valuation confirmed?
No. The $1.8–$2 trillion range comes from people familiar with Anthropic’s plans, via Bloomberg and CNBC. It is a reported valuation scenario — not a public filing, an offer price, or a completed deal.
Did Anthropic make a profit in 2025?
According to the reported prospectus figures, no: an operating loss of about $8 billion and a net loss of about $42 billion — though over $34 billion of the net loss is non-cash accounting adjustments. The company has claimed a positive adjusted operating income in Q2 2026.
When will the public see the real numbers?
When the confidential S-1 goes public — expected around the formal marketing push reported to start the week of 9 November, though Anthropic has not confirmed the timeline.
Sources and methodology
Sources: CNBC (13 Oct 2026) via alfw.ai; Bloomberg (1 Oct 2026) via AI Affairs and The Hindu BusinessLine; Reuters prospectus review (late Sep 2026) via NeoTeo and Motley Fool; Invezz; XTB’s Anthropic IPO guide. Method: desk research via web search on 14 October 2026 — no hands-on testing, no company confirmation (Anthropic declined to comment). Valuation and timeline figures are reported, not confirmed. This article is investor education, not financial advice. Verified 2026-10-14.
Govind Dheda covers AI tools, guides and prompts for OpenAIMaster — what is new, what is worth using, and how to put AI to work.
Feel free to email us at contact@openaimaster.ai — we are happy to help!


