📰 Key Takeaways

After Anthropic goes public, VCs like Menlo Ventures, Sequoia, Sound Ventures, and Spark Capital are expected to make billions. But according to the S-1 filing, none of their stakes reach the 5% threshold for principal shareholder disclosure, which is quite rare for a tech IPO. The reason is Anthropic’s private valuation of $965 billion, where 1% of the company is worth nearly $10 billion, so Menlo’s roughly $14 billion position is only about 1.5% of the company. The really big stakes sit with strategic investors who are also business partners, and Amazon is likely the biggest financial winner of this IPO.

Since committing up to $4 billion in 2023, Amazon has kept adding to its investment, mostly through convertible notes, a type of debt that converts into shares at an IPO or sale. As of June 30, Amazon held notes with $6.2 billion in principal, which is all of Anthropic’s debt. But as the valuation climbed, Amazon’s own financial statements valued those notes at $97.9 billion. As a result, Anthropic has to recognize fair value changes on its income statement, and the related item reached $72.4 billion in the first half of 2026, a non-cash charge. The notes can convert into up to 1.84 billion shares, but that’s capped at 12% of the “partially diluted” share count. Amazon can choose to waive the cap, and whatever can’t be converted rolls over to later quarters.

On top of that, Amazon put in $10 billion in non-voting preferred stock in Q2, which was $5 billion in each of the last two funding rounds. Together with the $8 billion it put into notes from 2023 to 2025, its total investment comes to about $18 billion. Amazon also keeps profiting from Anthropic’s commitment to spend at least $110 billion on AWS over the next 10 years.

💬 JudyAI Lab Take

Anthropic’s S-1 reveals something rare: VCs like Menlo and Sequoia each hold less than the 5% principal shareholder threshold, and the real big winner is Amazon, which is also a business partner.

This case shows a very clear capital structure in the AI industry: the compute costs of training and inference are so high that the biggest backers are often the suppliers too. Amazon has put in about $18 billion through convertible notes and non-voting preferred stock, and it also keeps earning from Anthropic’s commitment to spend at least $110 billion on AWS over the next 10 years, so the money flows in a closed loop. Another detail worth noting: the notes are valued at $97.9 billion on Amazon’s books, while Anthropic recognized a $72.4 billion non-cash charge from fair value changes in the first half of 2026, so the book numbers swing wildly with valuation. For AI builders, this means model vendors’ finances are deeply tied to their cloud providers, and picking an underlying model means picking a whole supply chain.

I’d suggest spending 10 minutes to list which models and cloud services your product relies on, and think about whether you have a workable backup plan if one of them changes its pricing or strategy.

📅 Original Article Info

🔗 Further Reading