Amazon Positioned to Reap Nearly $98 Billion From Anthropic's Listing

The Unusual Cap Table

When Anthropic filed its registration statement for a public listing, venture capital firms braced for billion-dollar paydays. Menlo Ventures, Sequoia, Sound Ventures, and Spark Capital each hold positions expected to return multiples on their early bets. Yet none of these firms appears in the principal stockholders table - the section of the filing reserved for investors holding at least 5 per cent of the company.

That absence is striking. In most technology listings, lead venture investors command double-digit percentage stakes. At Anthropic, the maths work differently. At a $965 billion private valuation, a single percentage point is worth nearly $10 billion. Menlo's roughly $14 billion stake, one of the largest among venture firms, represents only about 1.5 per cent of the company, according to figures compiled by Bloomberg.

The disparity points to a structural shift in who profits most when frontier AI companies go public. At Anthropic, the largest shareholders are not venture funds but strategic partners - companies that also provide the cloud infrastructure on which the AI firm runs.

Amazon's Convertible Bet

Amazon first committed to Anthropic in 2023, pledging up to $4 billion. It has since deployed billions more, but much of that capital took an unconventional form: convertible notes, a type of debt that converts into equity upon a listing or sale.

As of 30 June, Amazon held convertible notes with $6.2 billion in principal outstanding. That figure represents Anthropic's entire debt load. But as Anthropic's valuation climbed across successive funding rounds, the shares those notes will convert into became far more valuable. Amazon's own quarterly filings valued the note holdings at $97.9 billion as of the same date.

The accounting mechanics are unusual. As the fair value of Amazon's notes surged, Anthropic was required to record the growing liability on its income statement. For the first half of 2026, the line item "change in fair value of convertible notes, derivative liabilities, and SAFEs" totalled $72.4 billion. The charge is non-cash; no money left the company. But it reflects the scale of Amazon's paper gain.

The notes could convert into as many as 1.84 billion shares, according to the registration statement. A cap limits Amazon to 12 per cent of Anthropic's "partially diluted" share count, a figure that includes certain unissued shares such as stock options. Amazon can waive the cap. Any notes that cannot convert immediately will remain outstanding and convert in later quarters, subject to the same limit.

Equity on Top

The convertible notes are only part of Amazon's position. In the second quarter, the company invested an additional $10 billion in non-voting preferred stock, according to Amazon's quarterly report for the period ended 30 June. That equity investment came in two tranches of $5 billion each, deployed across Anthropic's last two funding rounds.

Combined with the $8 billion Amazon invested in convertible notes from 2023 through 2025, the total outlay stands at roughly $18 billion. The return on that investment, measured at the June mark, exceeds $97 billion - a multiple of more than five times in under three years.

Beyond its equity gains, Amazon has locked in revenue from Anthropic's operations. The AI company has committed to spend at least $110 billion on Amazon Web Services over the next decade, a cloud contract that will generate recurring income independent of any share price movement.

Google's Parallel Play

Google has pursued a similar strategy. Court filings reported by The New York Times indicated that the search giant held about 14 per cent of Anthropic as of early last year. It has invested more since, including a $10 billion commitment in April. Subsequent funding rounds likely diluted that stake; PitchBook analyst Harrison Rolfes estimates Google's current ownership at closer to 5.6 per cent.

Google has also secured at least $114 billion in cloud commitments from Anthropic through 2033. Like Amazon, it stands to profit both from the listing and from the infrastructure spend that follows.

Strategic Capital Versus Venture Capital

The structure of Anthropic's cap table reflects a broader pattern emerging in frontier AI. Companies building large language models require extraordinary amounts of capital - not only for research and engineering but for compute. Training runs for state-of-the-art models can cost hundreds of millions of dollars; inference at scale adds further expense.

Cloud providers can supply that compute, and they can supply capital. In exchange, they secure long-term revenue contracts and equity stakes that, in Anthropic's case, exceed the holdings of any venture firm by an order of magnitude.

At Opentechwire, we have tracked similar arrangements across the sector. Microsoft's partnership with OpenAI, which includes a reported $13 billion investment and a multiyear Azure compute commitment, follows a comparable logic. Inflection AI sold much of its team to Microsoft in a deal that blurred the lines between acqui-hire and cloud partnership. In each case, the strategic investor gains influence over model development, access to cutting-edge capabilities, and a hedge against disruption - all while locking in infrastructure spend.

The venture firms that backed Anthropic early will still earn substantial returns. A 1 per cent stake in a company valued at nearly $1 trillion is a lucrative outcome by any measure. But the returns are dwarfed by those accruing to Amazon and Google, whose convertible structures and equity positions were sized to match the scale of their cloud commitments.

The Risk of Dependence

Anthropic's registration statement includes a disclosure that underscores the tension inherent in these arrangements. It warns that some of the cloud providers on which the company depends also build rival AI models. Their interests may not always align with Anthropic's.

Amazon, for instance, has developed its own Titan family of models and offers third-party models through Bedrock, its managed AI service. Google competes directly with Anthropic through its Gemini model family. Both companies have an interest in ensuring that Anthropic remains a viable competitor to OpenAI and others, thereby sustaining demand for their cloud platforms. But both also have an interest in promoting their own models.

The dynamic creates a form of structural dependency. Anthropic relies on AWS and Google Cloud for the compute required to train and serve its models. It has committed more than $220 billion in cloud spend over the next decade, a figure that exceeds the total capital it has raised. If either provider were to restrict access, raise prices, or prioritise its own models, Anthropic's ability to operate at scale would be compromised.

The registration statement notes that the company is working to diversify its infrastructure. It has begun designing its own chips in partnership with Broadcom and TSMC, and it is exploring alternative cloud providers. But those efforts will take years to bear fruit. In the near term, Anthropic remains tightly coupled to the two companies that also happen to be its largest shareholders.

Implications for the Next Wave

The Anthropic listing offers a preview of how value will be distributed as other frontier AI companies go public. Venture capital will continue to play a role in early-stage funding, but the largest financial outcomes may increasingly flow to strategic investors who can supply both capital and compute at the scale required to compete.

That shift has consequences for competition and governance. When a company's largest shareholders are also its infrastructure providers - and also its competitors - the incentives become complex. Amazon and Google have a financial interest in Anthropic's success, but they also have an interest in ensuring that success does not come at the expense of their own AI ambitions.

For founders and early employees, the shift means that ownership will be more diluted by the time a company reaches a listing. For venture firms, it means that even successful investments may yield smaller percentage stakes than in previous technology cycles. And for regulators, it raises questions about vertical integration and the concentration of power in the hands of a small number of cloud providers.

At a $965 billion valuation, Anthropic is one of the most valuable private companies ever to file for a public listing. The returns it generates will be measured in tens of billions of dollars. But the distribution of those returns tells a story about who holds leverage in the AI economy - and it is not the venture firms that wrote the first cheques.