
Anthropic confirmed on Tuesday, November 10, that it will price its initial public offering at $70.00 to $78.00 per share, a range that implies roughly 17.9 billion to 20 billion shares outstanding against a $1.4 trillion target valuation, with trading expected on Thursday or Friday of next week (November 19 or 20). The band sits below earlier, more ambitious expectations yet still ranks the deal among the largest listings in Wall Street history. The core conclusion here is that Anthropic is
The hard facts line up along a short timeline. On Tuesday, November 10, Anthropic fixed its offering range at $70.00 to $78.00 per share in an amended registration filing and confirmed the numbers publicly. The band spans $8, or 11.4 percent of the low end, with a $74 midpoint that leaves the customary room for book-building adjustments. The Wall Street Journal, citing people familiar with the matter, reported that the company wants shares to begin trading "Thursday or Friday next [week]" — November 19 or 20 — and Wall Street convention puts final pricing the evening before, which points to November 18. The company acknowledged the deal will rank among the biggest IPOs ever while still falling short of prior expectations. Dividing the $1.4 trillion target valuation by the range bounds implies roughly 17.9 billion to 20 billion shares, so every $1 move in the offer price shifts total
Why now comes down to a window rather than a peak. In 2026 the capital intensity of AI training and inference has pushed single-round private financings into the hundreds of billions, straining what private markets can absorb; listing has moved from optional to necessary. Comparables now exist: Arm's September 2023 listing proved that mega-cap technology names can price into a soft tape, and CoreWeave's March 2025 debut gave public investors a direct read on AI infrastructure. Anthropic's own path is well documented — public reporting puts its September 2025 Series F at $13 billion raised on a $183 billion valuation, after which expectations were pushed toward the trillion-dollar mark,
Liu Gong's take: this is an IPO that sprints for a window, not one that cashes in on a peak. Within 12 months, by November 2027, Anthropic's market capitalization will most likely trade between $0.8 trillion and $1.2 trillion and will struggle to hold $1.4 trillion, because public markets price revenue multiples, not narrative.
From a technical standpoint, what listing really changes is cost transparency. Compute for training and inference is Anthropic's largest single expense line, and once it is reported quarterly, compute cost per dollar of revenue turns from a trade secret into a public metric — which constrains model cadence in reverse. A lab could once spend freely for a capability jump; a public lab must answer a margin question before every major release. In industrial inspection work, Liu Gong has seen the blunter version of the same logic: clients switch suppliers for a 5 percent cut in inference cost, because the cost curve, not a leaderboard, is the source of pricing power on the enterprise side. One detail is easy to miss: on public parameters, each $1 of offer price equals roughly $180 million to $200 million of valuation, so the $8 band hands underwriters about $1.5 billion of valuation flexibility — small order-book shifts during the roadshow get amplified into real pricing concessions.
On the business side, scarcity premium can carry a first day, but not a fourth earnings season. Against the $183 billion valuation of the September 2025 Series F, $1.4 trillion is a cross-market re-rating of roughly 7.6 times. Moves of that size normally need two to three quarters of reported results to validate, while an IPO compresses validation into a two-week roadshow. Institutions are buying the scarcity of the first pure AI model listing; the eventual sell case will be revenue growth racing compute spend.
For the industry, the chain reaction depends on the first 30 trading sessions. A clean raise above $70 billion re-anchors private negotiations across the sector to new multiples; a broken first day most likely pushes the AI candidates queuing for the first half of 2027 back by three to six months. Liu Gong's prediction is a checkable signal: within six months, by May 2027, at least one more frontier AI lab or compute supplier will file for a US listing, and its range will visibly reference the discount applied here.
- Offer range of $70.00 to $78.00 per share, an $8 spread equal to 11.4 percent of the low end, with a $74 midpoint.
- Divided by the $1.4 trillion target valuation, the range implies about 17.9 billion shares at $78 and 20 billion at $70, so each $1 of offer price equals roughly $180 million to $200 million of valuation.
- At a conventional 5 to 10 percent float, gross proceeds would land near $70 billion to $140 billion, or 2.4 to 4.8 times Saudi Aramco's $29.4 billion record from
- – Versus the $183 billion valuation of the September 2025 Series F, $1.4 trillion represents a 7.6-fold cross-market re-rating; against Arm's roughly $54.5 billion listing in 2023 and CoreWeave's roughly
For the industry, the three-month effect is a reset reference price: private negotiations will treat the Anthropic band as the new benchmark, and valuation talks across the sector will lengthen. Within a year, if the company holds a trillion-dollar valuation, AI will have its first pure-play index candidate, bringing structural passive demand — and turning AI earnings season into a volatility source for the whole technology sector.
For ordinary users, nothing much changes in three months, but the odds of adjustments to free tiers and API pricing rise, since public markets demand visible revenue quality and generous free access is the easiest line to trim. Within a year, product weight
Market reaction clusters along three lines. On the bullish side the argument is scarcity: this is the first opportunity for public investors to own a pure large-model company outright, and allocation demand plus index inclusion expectations form a near-term bid, with sell-side research broadly framing $1.4 trillion as a discounted entry point. Skeptics focus on the size of the cross-market re-rating — roughly 7.6 times from $183 billion to $1.4 trillion without audited results to validate it — and some hedge fund commentators argue the $70 floor is where real demand is being tested. Developers care about something more concrete: API pricing, service-level commitments and whether release
Questions worth exploring:
- Use the roughly $200 million per $1 of offer price to explain how underwriters manage order-book risk through range width in mega-listings.
- Compare the first 90 trading days of Arm in 2023 and CoreWeave in 2025 to model pricing efficiency for pure AI listings.
- Examine how mandatory disclosure of inference cost as a share of revenue will reshape model iteration strategy at large labs.
Further reading: 2026 Frontier LLM Comparison: A Practical Guide — a durable, evergreen reference. Bookmark it.
This item comes from The Verge AI (published 2026-11-10 14:06:54). Wenfeng Radar provides a Chinese/English summary and commentary on overseas AI developments; the original content is owned by its author.
Daily aggregation of overseas AI developments with in-depth analysis. Bookmark us to never miss an important signal; back to homepage.
