Anthropic's $64 Billion IPO Draws Six Times Oversubscription in Three Days

Anthropic's $64 billion initial public offering was six times oversubscribed by the third day of its roadshow, Reuters reported on November 13, 2026. The demand outpaced the early days of SpaceX's roadshow, according to people familiar with the offering.

The numbers came fast. Orders reached about twice the shares available by the end of day one. Two days later, the book stood at six times oversubscription. Bankers and investors called the early demand impressive for an offering of this size, and they said they expected the order book to grow substantially.

The people familiar with the process cautioned that demand and other IPO details could still change. Anthropic declined to comment on the reported demand.

What are the offering's terms?

Anthropic launched the roadshow on November 10, 2026. The company is offering 825 million Class A shares at an expected price of $70 to $78 per share. Underwriters will hold a 30-day option to buy up to 123.75 million additional shares.

The roadshow runs through November 18, 2026, when the $64 billion deal is expected to price. Anthropic and its investment banks are meeting prospective investors in New York, Boston and San Francisco to gauge demand.

Anthropic has applied to list on the Nasdaq Global Select Market. The offering's registration had not yet become effective when the terms were announced.

The takeaway: a deal of this size, priced in a single week, with a shares-plus-greenshoe structure that could raise the total further.

Why are investors hesitant despite the demand?

The oversubscription tells one story. The roadshow meetings tell a more cautious one.

Reuters reported that investors are testing two parts of Anthropic's investment case: whether the company can stay ahead in advanced AI models as Alphabet's Google and other rivals close the gap, and whether Claude can expand from coding into broader knowledge-work uses.

The debate extends further. Investors are asking how many companies the AI market can sustain at the top, and whether comparable rival models could erode Anthropic's premium pricing. Claude currently commands premium prices, and the valuation case depends on that holding.

Disclosure has added to the caution. One investor who attended a roadshow meeting said Anthropic had not provided detailed answers to some questions. That limited disclosure has made some investors wary of a high-valuation offering during a broader shakeout in the AI sector.

The format of the meetings themselves drew comment. One attendee said a group meeting lacked a conference call for institutional investors to dial in and ask questions — a departure from standard practice. The attendee said Goldman Sachs and Morgan Stanley representatives largely read prepared remarks rather than solicit questions. Morgan Stanley declined to comment. Goldman Sachs did not immediately respond.

The takeaway: strong demand and real doubts are coexisting in the same order book.

How do the revenue models work?

The valuation debate hinges on 2028. Banks are anchoring their models to projected 2028 revenue of approximately $270 billion to $280 billion, according to two people familiar with the projections. The models apply enterprise-value-to-revenue multiples — a standard method that compares a company's value with its expected sales — to arrive at valuation estimates.

The revenue assumptions underneath those models vary widely:

• Sell-side projections: roughly $52 billion to $53 billion in 2026 revenue and $135 billion to $137 billion in 2027, according to one person familiar with the projections.
• Buy-side forecasts: one investor who placed an order in the IPO said certain buy-side forecasts were 10% to 15% higher for those years.
• Bear case: an investor who conducted due diligence but decided not to participate used a 2028 outlook about 30% below the sell-side estimate.

That spread illustrates the core problem. Two investors looking at the same company can arrive at materially different 2028 revenue pictures — and the valuation multiples sit on top of those numbers, compounding the disagreement.

The takeaway: when a 30% gap in revenue assumptions separates participants from non-participants, the model is doing a lot of the work.

What are the valuation benchmarks?

SpaceX and the cybersecurity company CrowdStrike are among the valuation benchmarks, one person familiar with the projections said. There is a difference in how the models treat them. The valuation models for Anthropic focus on 2028 revenue. SpaceX's case relied on projected 2029 revenue — and some investors viewed those projections as particularly aggressive.

The comparison matters because it frames what investors will accept. If SpaceX's forward-revenue anchoring struck some as aggressive, Anthropic's one-year-closer horizon may read as more defensible. But the same structural bet remains: paying today for revenue that exists mostly in a spreadsheet.

The takeaway: benchmarking against SpaceX means inheriting the argument over how far forward a valuation can credibly reach.

What happens on November 18?

The early subscription figures establish strong demand for the shares. The unresolved question is valuation — specifically, whether Claude can maintain its competitive position and premium pricing strongly enough to support the revenue assumptions banks are using through 2028.

Google's closing gap is the pressure point. If comparable rival models keep arriving at lower prices, Anthropic's premium pricing erodes, and with it the revenue trajectory the entire valuation rests on. Investors placing orders now are effectively betting that Claude's edge in advanced models — and its expansion beyond coding into broader knowledge work — outlasts the competition.

The pricing on November 18 will show where that bet settles. But the real test arrives later, in the quarterly results that either confirm or break the 2028 models.