Amazon Holds 15 Per Cent of Anthropic at IPO, Could Have Taken 24
The cloud giant converted only part of its promissory notes into equity, leaving billions in optionality as the AI firm goes public with Google owning another 14 per cent.
A Stake Built on Early Conviction
Anthropic disclosed the structure of its largest investment agreements on Tuesday, revealing that Amazon will hold 14.7 per cent of the AI company at the time of its initial public offering. The stake comes from the partial conversion of promissory notes Amazon obtained when it invested billions in Anthropic across 2023 and 2024, a period when the startup's Claude models had yet to gain significant enterprise traction.
The notes carry conversion prices of about $2.09 or $5.05 per share, depending on the year Amazon made the investment. Amazon has chosen to convert a portion of those notes into a special class of non-voting shares that will deliver dividends and can convert to standard voting Class A shares upon sale. The remainder of the notes, worth billions of dollars, Amazon has opted to hold.
Had Amazon converted the full value of its notes at the time of the IPO, its ownership would have reached 23.8 per cent of Anthropic. The decision to leave a substantial portion unconverted preserves optionality for the cloud provider as Anthropic's valuation continues to evolve in the public markets.
Google's Position and the Private Placement
Google will own 13.7 per cent of Anthropic, also structured as non-voting shares. That stake is derived from both the conversion of promissory notes and participation in a private placement conducted alongside the IPO.
The private placement attracted a consortium of seven major technology firms: AMD, Broadcom, Google, Micron, Microsoft, Nvidia, and TSMC Partners. Together, the group agreed to purchase $26 billion of non-voting stock. Google and Nvidia each committed $10 billion to the placement, making them the largest individual buyers in that tranche.
The structure ensures that Anthropic secures substantial capital from strategic partners without diluting voting control, a common arrangement in IPOs where founders and early backers wish to retain governance rights even as the shareholder base expands.
The Cloud Commitment Underneath
Amazon's financial return extends beyond equity appreciation. Anthropic has committed to spending more than $100 billion on Amazon Web Services over the next decade, a contract that will generate recurring revenue for AWS independent of the IPO proceeds or share price performance.
That commitment was negotiated as part of the original investment agreements and reflects the infrastructure intensity of training and deploying large language models at scale. For Amazon, the arrangement layers a predictable revenue stream atop the equity upside, hedging the investment across two distinct business lines.
The cloud spend commitment also ties Anthropic to AWS infrastructure for the foreseeable future, limiting the company's flexibility to shift workloads to competing platforms such as Google Cloud or Microsoft Azure. In that sense, the investment functions as both a financial instrument and a strategic lock-in.
A Bet Placed Before the Breakout
When Amazon first invested in Anthropic, the company occupied a distant position behind OpenAI in both market perception and product adoption. Claude had not yet achieved the enterprise momentum that would later follow the release of Claude Code, and the broader narrative around foundation models was still dominated by GPT-4 and its derivatives.
Amazon's willingness to deploy billions at conversion prices near $2 and $5 per share reflects a bet placed before the market had priced in Anthropic's eventual competitive standing. The decision to hold a portion of the notes unconverted suggests that Amazon views the current IPO valuation as a floor rather than a ceiling, preserving the ability to increase its stake if Anthropic's public market performance justifies further conversion.
Google's participation, both through note conversion and the private placement, signals a similar calculus. The search and cloud giant has its own AI ambitions through DeepMind and Gemini, yet it has chosen to allocate capital to a third-party model provider, likely as a hedge and as a means of securing access to Anthropic's research and product roadmap.
What the Ownership Structure Reveals
The combined 28.4 per cent stake held by Amazon and Google at IPO positions the two cloud providers as Anthropic's largest institutional shareholders. Neither holds voting rights through these shares, but both wield influence through commercial agreements, infrastructure dependencies, and board representation that typically accompanies investments of this scale.
The decision to issue non-voting shares to strategic investors is a deliberate governance choice. It allows Anthropic's founders and early venture backers to retain control over corporate decisions while still raising capital from parties whose interests may not always align with those of the company's leadership.
For Amazon, the structure also offers a path to increase its stake over time without triggering immediate dilution or governance complications. If the company converts the remaining notes in tranches, it can adjust its ownership in response to Anthropic's performance and its own strategic priorities.
The IPO marks a turning point for Anthropic, but the ownership structure disclosed this week shows that the company's largest backers placed their bets years earlier, at valuations that now appear prescient. Amazon's decision to leave billions in optionality unconverted suggests that the story is far from finished.



