Amazon invests $50 billion in OpenAI as part of $110 billion funding round

OpenAI announced on February 27 what it described as $110 billion in new investment at a $730 billion pre-money valuation.^[1] The round consists of $50 billion from Amazon, $30 billion from SoftBank, and $30 billion from NVIDIA.^[1] Additional investors are expected to join as the round progresses.^[1]

Alongside the capital raise, Amazon and OpenAI announced a multi-year strategic partnership that includes joint development of a new "Stateful Runtime Environment" on Amazon Bedrock, exclusive third-party cloud distribution of OpenAI's enterprise platform Frontier through AWS, and a commitment by OpenAI to consume approximately 2 gigawatts of AWS Trainium chip capacity.^[2]

The deal significantly expands OpenAI's cloud infrastructure footprint beyond its original exclusive relationship with Microsoft, though both companies issued a joint statement the same day affirming that their partnership "remains strong and central."^[3]

Investment structure

Amazon's $50 billion is structured in two phases. The company will invest an initial $15 billion, followed by an additional $35 billion "in the coming months when certain conditions are met."^[2] The official announcement does not specify what those conditions are. The Information reported on February 25 that the additional $35 billion may be contingent on OpenAI either completing an initial public offering or reaching an artificial general intelligence milestone.^[4] TechCrunch noted that the announcement "confirms the funding split" but uses vaguer language than what The Information described.^[5]

SoftBank’s $30 billion will be invested through SoftBank Vision Fund 2 in three equal $10 billion tranches, scheduled for April 1, July 1, and October 1, 2026, according to its press filing. The shares are preferred stock that automatically convert into common shares upon an IPO or related listing transaction. ^[6] SoftBank disclosed that the investment will bring its cumulative stake in OpenAI to approximately $64.6 billion, representing about 13% ownership.^[6] The SoftBank board authorized the investment on February 20, and the agreement was signed on February 27 (U.S. time).^[6] SoftBank stated the investment will be initially financed through bridge loans and other financing arrangements.^[6]

NVIDIA's $30 billion comes alongside a commitment by OpenAI to use 3 gigawatts of dedicated inference capacity and 2 gigawatts of training on NVIDIA's Vera Rubin systems.^[1] OpenAI said this builds on existing Hopper and Blackwell deployments across Microsoft, OCI, and CoreWeave.^[1]

It is worth noting that a significant portion of the dollar amounts in AI funding rounds of this kind may take the form of cloud credits or infrastructure commitments rather than cash. TechCrunch noted that "it is likely that a significant portion of the dollar amount comes in the form of services rather than cash, although the precise split was not disclosed."^[5]

What the Amazon partnership covers

The partnership has four main components, according to the joint press release.^[2]

Stateful Runtime Environment. Amazon and OpenAI are jointly developing a "Stateful Runtime Environment" powered by OpenAI models, to be available through Amazon Bedrock.^[2] OpenAI describes stateful developer environments as allowing models to "keep context, remember prior work, work across software tools and data sources, and access compute."^[2] This is distinct from conventional stateless API calls, where each request is processed independently. The environment is expected to launch "in the next few months."^[2]

The stateful-versus-stateless distinction matters because of Microsoft's existing rights. In their joint statement, Microsoft and OpenAI clarified that Azure remains the exclusive cloud provider for stateless OpenAI APIs, and that "any stateless API calls to OpenAI models that result from a collaboration between OpenAI and any third party — including Amazon — would be hosted on Azure."^[3] The Stateful Runtime Environment appears to represent a new deployment category that falls outside that exclusivity. This interpretation is based on the language of the Microsoft joint statement, but has not been explicitly confirmed by any party.

Frontier Distribution. AWS will be the "exclusive third-party cloud distribution provider" for OpenAI Frontier, the company's enterprise platform for building, deploying, and managing teams of AI agents.^[2] Microsoft's joint statement clarified that "OpenAI's first party products, including Frontier, will continue to be hosted on Azure."^[3] The distinction appears to be between hosting (Azure) and third-party cloud distribution (AWS).

Trainium Compute. The partnership expands on an existing $38 billion multi-year AWS agreement by an additional $100 billion over eight years.^[2] OpenAI commits to consuming approximately 2 gigawatts of Trainium capacity, spanning both Trainium3 and the next-generation Trainium4 chips.^[2] Trainium4 is expected to begin delivery in 2027 and will feature higher FP4 compute performance, expanded memory bandwidth, and increased high-bandwidth memory capacity, according to the press release.^[2]

In his post on X announcing the deal, Amazon CEO Andy Jassy described Trainium chips as "30-40% more price performant than comparable GPUs" and noted that "both the leading AI labs now have made significant commitments to Trainium."^[7] The reference to "both" leading AI labs appears to be a reference to Anthropic, which also uses Trainium through its AWS partnership, though Jassy did not name the other company. The "30-40%" performance figure is Jassy's characterization; independent benchmarks comparing Trainium3 to current-generation NVIDIA GPUs on equivalent workloads were not cited.

Custom Models for Amazon. OpenAI and Amazon will develop customized models for Amazon's developers to use in consumer-facing applications.^[2] These would complement Amazon's existing Nova model family.^[2] CNBC reported earlier in February that OpenAI models could help power Alexa, Amazon's digital assistant, as part of the investment discussions.^[8] Alexa+ currently routes many complex queries through Anthropic's Claude, though Amazon has described the service as "model agnostic" and said it uses more than 70 different models.^[8]

How this affects the Microsoft-OpenAI relationship

Microsoft and OpenAI published a joint statement on February 27 addressing the Amazon partnership directly.^[3] The statement said that Microsoft retains its "exclusive license and access to intellectual property across OpenAI models and products," that the revenue-sharing arrangement is unchanged, and that Azure remains the exclusive cloud provider for stateless OpenAI APIs.^[3] Any stateless API calls resulting from third-party collaborations, including with Amazon, would be hosted on Azure.^[3] The contractual definition of AGI and the process for determining whether it has been achieved also remain the same, according to the statement.^[3] Microsoft stated that collaborations like the Amazon partnership "were always contemplated under our agreements."^[3]

OpenAI separately told CNBC that nothing about the announcement "in any way changes the terms" of its Microsoft partnership.^[9]

The broader trajectory, however, shows a steady loosening of OpenAI's cloud exclusivity with Microsoft. Microsoft was OpenAI's exclusive cloud provider until January 2025, when the arrangement shifted to a "right of first refusal" model as part of the Stargate Project announcement.^[10] In October 2025, when OpenAI completed its conversion to a public benefit corporation, even that right of first refusal was removed. Under the definitive agreement signed that month, OpenAI committed to purchasing an incremental $250 billion in Azure services, but is now free to contract with any cloud provider without offering Microsoft first right.^[14] OpenAI subsequently signed cloud agreements with Oracle, Google, and AWS. The initial AWS deal, a $38 billion seven-year agreement announced in November 2025, was the first major infrastructure contract between the two companies.^[10] Today's announcement substantially deepens that relationship.

CNBC reported, citing unnamed sources, that OpenAI is now targeting approximately $600 billion in total compute spending by 2030, down from earlier projections of $1.4 trillion in infrastructure commitments.^[9]

Amazon's position: backing both OpenAI and Anthropic

Amazon has now made substantial investments in the two leading frontier AI labs. It previously invested $8 billion in Anthropic, for which AWS serves as the primary cloud and training provider.^[11] Amazon opened an $11 billion data center campus in Indiana designed to run Anthropic models and built the "Rainier" supercomputer using nearly 500,000 Trainium2 chips for Anthropic's workloads.^[5]

Adding a $50 billion investment in OpenAI positions Amazon as a major financial backer on both sides of the frontier model competition. Both companies' models are already available through Amazon Bedrock, AWS's managed AI service.

This dual approach ensures AWS has deep relationships with the leading AI labs regardless of competitive outcomes. It also raises questions — flagged by analysts and trade press before today's announcement — about how Amazon will manage data sharing, model exclusivity, and potential conflicts between two directly competing portfolio companies.^[12] Those questions are not addressed in any of the official announcements.

OpenAI growth metrics

OpenAI shared several self-reported usage figures alongside the announcement:^[1]

• More than 900 million weekly active users of ChatGPT

• More than 50 million consumer subscribers

• More than 9 million paying business users

• 1.6 million weekly Codex users, which OpenAI described as more than tripling since the start of the year

• January and February 2026 on track to be the largest months for new subscriber acquisition in the company's history

These figures are self-reported by OpenAI and have not been independently audited. For context, OpenAI reported 400 million weekly active users in February 2025, meaning the 900 million figure implies roughly 125% growth in one year.

CNBC separately reported, citing unnamed sources, that OpenAI projects total revenue of more than $280 billion by 2030, with roughly equal contributions from consumer and enterprise businesses.^[9] OpenAI has not publicly confirmed that projection.

What remains unclear

Several aspects of the deal are not fully explained by the official announcements.

The specific conditions that trigger Amazon's additional $35 billion investment are not disclosed in any of the three companies' official statements.

The precise split between cash investment and infrastructure credits or compute commitments within the $110 billion total has not been disclosed.^[5]

How the Stateful Runtime Environment relates to Microsoft's Azure exclusivity in practice — and whether this framework could be extended to other cloud providers beyond AWS — is not addressed in detail.

The operational mechanics of Amazon using customized OpenAI models alongside Anthropic's Claude in products like Alexa have not been elaborated beyond general statements about model-agnostic architecture.^[8]

Whether federal antitrust regulators will scrutinize the deal is an open question. Axios noted that while this is not a merger or acquisition, the FTC previously examined the relationship between OpenAI and Microsoft.^[13]