How OpenAI plans to win over the enterprise

Welcome to Runtime! Today: OpenAI's new chief revenue officer outlines a five-point plan for driving enterprise business, local opposition to new data-center construction is going to be a midterm campaign issue, and the latest funding rounds in enterprise tech.

Please forward this email to a friend or colleague! If it was forwarded to you, sign up here to get Runtime each week, and if you value independent enterprise tech journalism, click the button below and become a Runtime supporter today.

Get busy businessing

Even though it was bankrolled for years by one of the biggest enterprise software companies on the planet, most of OpenAI's business still comes from consumer-facing products like ChatGPT. Over the last year the company realized that situation needed to change if it planned on following through on the more than $1 trillion it has committed to spend on computing resources, and over the weekend it laid out a plan for competing in the enterprise.

OpenAI's Denise Dresser wrote a memo to employees Sunday obtained by The Verge that focuses on five ways that OpenAI can erase Anthropic's lead in the enterprise, where it is believed to have around 40% of the market for enterprise spending on large-language models, according to (Anthropic investor) Menlo Ventures. Dresser, recently hired as chief revenue officer after a long career at Salesforce, wrote that OpenAI's early advantage has eroded in part because these days "customers want fit: how well AI plugs into their workflows, knowledge, controls, and day-to-day operations, and how effectively it can be deployed, trusted, and improved over time."

• Dresser's first point was that OpenAI has more computing capacity than Anthropic, a shortage that continues to plague the company as demand for Claude Code exploded over the last six months.
• She referenced an upcoming model code-named "Spud," which "also delivers on everything that matters for high-value professional work: stronger reasoning, better understanding of intent and dependencies, better follow-through and more reliable output in production."
• Dresser also noted that OpenAI still has a chance to emerge as a leader in agent-management platforms, although it is competing with pretty much every single other enterprise software company to win that business.

Dresser made some interesting comments about OpenAI's new bestie: AWS, which was forced to embrace Anthropic and its models during the period in which Microsoft had exclusive access to OpenAI's models. AWS spent years fighting off accusations that it had missed the window on enterprise AI thanks to that deal, but Dresser suggested that Microsoft didn't take as much cloud infrastructure market share from AWS during that period as a lot of people might have thought.

• "Our Microsoft partnership has been foundational to our success. But it has also limited our ability to meet enterprises where they are – for many that’s Bedrock," she wrote.
• She called demand for the Amazon Stateful Runtime Environment introduced in February "staggering," and said "we are firing on all cylinders to establish this as a scaled distribution channel."
• Working with AWS gives OpenAI access to accounts that standardized on AWS long ago, she said, and a lot of those companies would have found it very difficult to switch model providers when Anthropic's models were readily available and extremely capable.

Dresser also argued that OpenAI needs to think of itself as "a platform company with multiple entry points and one integrated enterprise offering," and that it should aim to "own deployment" with its forward-deployed engineers. But it will take some work for OpenAI to shift from a consumer-oriented company to an enterprise-oriented one.

• Google Cloud went through this same process about seven years ago, transforming from a company that had a reputation for being very difficult to work with into a more focused organization under former Oracle executive Thomas Kurian, and it was not easy.
• OpenAI is under a lot more pressure to make this shift as well, given that the clock is ticking and it doesn't have the legendary advertising cash machine and stable, proven compute infrastructure that Google enjoyed during that transition.
• Access to multiple types of models might become very important to enterprise customers over time as agentic AI matures and different workloads need to tap into different model capabilities, which could make it harder for any one model provider to own the application layer.
• It also probably wouldn't hurt if OpenAI could manage to go a full year without turmoil in its executive ranks.

Byte me

As companies like Anthropic and OpenAI struggle to get more computing capacity online, they're running into more opposition in the U.S. than they probably expected a few years ago. Local communities prevented several data-center proposals from moving ahead over the last six months, and the trend only seems to be growing.

Voters in the town of Festus, Missouri, ousted four incumbent city council members by wide margins last week after the council approved a data-center construction plan last month, according to Politico. The residents of Port Washington, Wisconsin, also took action against data-center operators last week by passing a referendum giving them more control over how the town approves massive building projects.

In Wisconsin at least, the opposition also appears to be bipartisan, and there aren't a lot of issues that will unite Democrats and Republicans in 2026. But hopefully everyone can agree that while communities are allowed to make collective decisions about land use, violence against AI executives has no part in this debate.

Enterprise funding

Sygaldry raised $139 million in seed and Series A funding for quantum computers that will be designed to run alongside AI servers in data centers.

Bluefish landed $43 million in Series B funding for its marketing analysis software, which uses AI agents to help companies understand how LLMs are discussing their brands.

AfterQuery scored $30 million in Series A funding for its AI model training datasets, which promise to help AI companies build models centered around human expertise and sell them to companies that want to fire all their employees.

Prefix raised $7.5 million in seed funding for its facility-management software, which was designed for restaurants and retailers.

The Runtime roundup

Anthropic will start charging big customers on a per-use basis as it tries to manage its computing crunch, according to The Information.

Microsoft will take a portion of the computing capacity that Nscale was reserving for OpenAI until it began to pull back on Project Stargate, according to Bloomberg.

Thanks for reading — see you Thursday!