• by Laurie Sullivan , Yesterday

OpenAI finalized a $110 billion record-breaking funding round fueling the ChatGPT maker’s push to secure more computing power and talent for AI development, It also secured a partnership with one the largest marketplaces worldwide. 

Amazon's $50 billion investment in OpenAI's financing round announced Friday. The company will begin with an initial $15 billion investment, followed by another $35 billion in the coming months when certain conditions are met.

Separately, SoftBank Group, and Nvidia each contributed a separate $30 billion in funding, OpenAI said today.

As a longtime Anthropic financial backer, the funding round from Amazon tightens its relationship with OpenAI. As part of the agreement, OpenAI will use Amazon’s line of in-house AI chips, called Trainium.

OpenAI will spend an additional $100 billion on Amazon Web Services during the next eight years. The two companies in November announced a deal under which the model builder would use some $38 billion in AWS services over seven years.

The collaboration with Amazon means working with Amazon Web Services companies to deliver what OpenAI called a new "Stateful Runtime Environment" that runs natively in Amazon Bedrock, powered by its technology. AWS customers will have access to OpenAI intelligence through the Runtime Environment, optimized for AWS infrastructure for agentic workflows.

In the context of a stateful environment, the partnership will change how advertisers build and deploy AI-driven campaigns by creating a persistent "brain" for their digital tools.

These models will have access to run parts of the workflow that make computer systems process data and assign tasks like compute, memory and identity, Amazon wrote in a blog post.

In other words, a Stateful Runtime Environment allows developers to retain context from tasks, remember prior work, work across software tools and data sources, and access compute. They're designed to handle ongoing projects and workflows, Amazon explained.

Instead of restarting an AI agent's logic each time a user clicks or a data point changes, a stateful environment allows the agent to remember prior work and carry through to the next task. This is critical for long-running tasks like multi-step campaign optimization or evolving customer support interactions.

Many prototype agents are based on stateless APIs that can tackle simple tasks like one prompt, one answer, and maybe one tool call, OpenAI explained in a blog post. It called out production work like real workflows as being different because those require several steps, context from previous actions, approvals, and system state. They also require guardrails in secure environments.

Stateless APIs require building the supporting orchestration layer for agents. The burden is on development teams to figure out how state is stored, how tools are invoked, how errors are handled, and how long-running tasks resume safely.