Seaport offers preliminary thoughts on Anthropic stock
Seaport Research Partners has offered preliminary insights into Anthropic's upcoming initial public offering (IPO). The research firm anticipates the AI developer will aim to raise between $1.5 trillion and $2 trillion, assuming a valuation of $1.64 trillion based on discounted cash flow models. Anthropic, which transitioned from zero revenue to $2.2 billion in just 7 quarters, generated $11.55 billion in revenue during the second quarter of 2026.
The company's revenue per employee in the third quarter of 2026 reached $3.19 million, significantly outpacing major technology counterparts. With 40 million customers, including 11,900 spending over $100,000 annually and 250 exceeding $10 million, Anthropic boasts an impressive 900% customer retention rate. The firm projects a return to operating losses in 2027 as the company executes "Anthropic Direct," a massive infrastructure transition.
Anthropic has secured commitments totaling nearly $800 billion for purchasing and leasing data center hardware through 2033, including a $161.2 billion lease with Broadcom and $114.5 billion and $120.5 billion commitments to Google Cloud and SpaceXAI, respectively. To fund the transition, Anthropic raised an additional $122 billion in equity and convertible debt in 2026, bringing cumulative capital raised to over $141 billion.
However, the company faces challenges due to its unique corporate governance structure and intense competition. As a Public Benefit Corporation, Anthropic operates under a Long-Term Benefit Trust that holds special class shares granting it control over four of the seven board members. Post-IPO, public investors will hold just 49.9% of voting rights, effectively ceding control to the LTBT and co-founders.
Despite facing pressure from open-weight models with lower gross margins, Anthropic has maintained premium pricing power, listing its Opus 5 model at $10 per million tokens, above rivals.
Written by urgent.news from Investing.com's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.