When Anthropic makes its public market debut, plenty of investors will be cashing enormous checks. Firms like Menlo Ventures, Sequoia, Sound Ventures, and Spark Capital are sitting on stakes expected to be worth billions. But here's the twist: according to Anthropic's S-1 filing, not a single one of those venture firms holds 5% of the company—the cutoff for appearing on the principal stockholders' table. That's a striking departure from the norm. In a typical tech IPO, the lead venture backers of a startup hold double-digit percentage ownership.

Anthropic's sheer scale explains the anomaly. With a private valuation of $965 billion, even a 1% slice is worth close to $10 billion. Menlo's position, for instance, sits at roughly $14 billion—which amounts to only about 1.5% of the company, based on figures recently reported by Bloomberg.

EMBED_PLACEHOLDER_0

So while the venture firms may not hold big percentage stakes, Anthropic did sell substantial ownership to strategic partners—companies that double as its business allies. In fact, Amazon is positioned to be the single biggest financial winner from the IPO.

Amazon's relationship with Anthropic began in 2023, when it committed to invest as much as $4 billion in the AI company. Since then, it has poured in billions more. Rather than taking stock outright, Amazon structured much of its investment as convertible notes—debt that converts into shares when an IPO or sale occurs. Per the S-1, Amazon holds convertible notes with $6.2 billion in principal outstanding as of June 30. That accounts for Anthropic's entire debt load.

As Anthropic's valuation climbed through successive funding rounds, though, the shares those notes will convert into became dramatically more valuable. Amazon's own filings pegged its note holdings at $97.9 billion as of June 30. Because the value of those convertible notes kept surging, Anthropic had to record the growing liability on its books. It reported the adjustments on its income statement under "change in fair value of convertible notes, derivative liabilities, and SAFEs"—a line item that totaled $72.4 billion for the first half of 2026. The charge is non-cash, meaning no actual money left the company.

According to the S-1, Amazon's notes could convert into as many as 1.84 billion shares. However, a cap restricts Amazon to 12% of Anthropic's "partially diluted" share count—a measure that includes some shares not yet issued, such as stock options. Amazon has the option to waive that cap. Any notes that can't convert will remain outstanding and convert in later quarters, subject to the same limit.

The notes are only part of what Amazon holds. In the second quarter, it invested another $10 billion in nonvoting preferred stock (a form of equity), according to Amazon's quarterly report for the period ended June 30. It did so by putting $5 billion into each of Anthropic's last two funding rounds. Add that to the $8 billion it put into the notes from 2023 through 2025, and Amazon's total investment comes to roughly $18 billion.

Beyond stock gains, Amazon will keep generating substantial revenue from Anthropic, which has pledged to spend at least $110 billion on AWS over the next decade.

Google is also in line for a major payday—both from the listing and from its cloud computing arrangements with Anthropic. The search giant held about 14% of Anthropic as of early last year, according to court filings reported by The New York Times, and has invested more since, including $10 billion in April. (That 14% figure was pre-IPO, so the listing will dilute it to roughly 11.6%.) Google has also locked in at least $114 billion in cloud commitments from Anthropic through 2033.

EMBED_PLACEHOLDER_1

The venture capitalists poised for massive paydays have little to complain about, but the S-1 makes one thing clear: the biggest beneficiaries of Anthropic's IPO may well be tech giants like Amazon—companies that backed a would-be rival early on and now sell it much of the computing power it relies on. The filing itself cautions that some of those cloud providers also build competing AI models, and that their interests may not always align with Anthropic's.