TheChek News Flow

While OpenAI’s younger rival Anthropic is winning over business customers, OpenAI is trying to convince companies it’s more than a consumer chatbot.

The Takeaway

Some of OpenAI’s most attention-grabbing efforts over the past year have involved products for consumers, from a social app to an AI-powered device it plans to announce later this year. Last week, though, the company’s CEO, Sam Altman, gathered Disney CEO Bob Iger and other corporate executives in San Francisco to deliver a message: OpenAI is serious about catering to business customers.

Over a lavish multicourse dinner accompanied by fine wines, Altman told attendees OpenAI could be a one-stop shop for all their AI needs, from ChatGPT to its coding tool, Codex, to its models for automating workflows, said a person with knowledge of the company’s plans.

The gathering was intended to preview a new OpenAI offering aimed at large companies, said another person with knowledge of the company’s plans. It couldn’t be learned what that offering is, but it aims to help business customers with large AI transformations. That generally means overhauling their existing technology to incorporate AI into a variety of operations, from customer service to rewriting older application code to how companies organize their data.

The new offering also aims to unify businesses’ efforts to use AI. That could involve bundling OpenAI’s various offerings into one that can make it easier for corporate clients to track their spending. The company currently offers businesses different tiers of ChatGPT, or higher-priced features such as advanced models and unlimited messages, as well as access to its models through an application programming interface, among other products.

To some attendees of the dinner, the subtext was clear: Altman wanted to steer customers away from archrival Anthropic, whose coding and workplace automation tools, Claude Code and Cowork, have captured the business world’s attention over the past month.

Anthropic last year projected it would generate more money from selling AI models to businesses through an API than OpenAI, according to both companies’ financial disclosures. That’s a notable result, given that OpenAI had a head start of several years in that market.

That said, OpenAI’s enterprise business is likely bigger than Anthropic’s because many businesses pay for ChatGPT, in addition to paying for its AI models through an API. But Anthropic has gained a reputation among corporate executives as the go-to provider of AI for businesses, say some large customers. That’s in large part because it hasn’t aggressively pursued many new consumer features.

OpenAI, in contrast, has continued to introduce such products at a dizzying speed. And it seems to still be figuring out how to shape its strategy for businesses, some of these large customers say. For most of the past few years, it has benefited from individual users of ChatGPT to spread the word with employers, to which it then sells large subscription contracts. But selling larger contracts to corporate customers is a more drawn-out process, which typically involves multiple meetings over many months with different executives, who are used to being wined and dined.

It wants more of the clients, like Databricks and Intuit, that come directly to OpenAI to use its products, said a person with knowledge of the company’s enterprise plans.

During the past week during the World Economic Forum at Davos, Switzerland, two senior OpenAI executives—Chief Operating Officer Brad Lightcap and Chief Revenue Officer Denise Dresser, who previously was CEO of Slack—were also talking up potential corporate clients, said one of the people. Back in San Francisco, the company told employees on Wednesday that a recently recruited top AI researcher from Thinking Machines Lab, Barret Zoph, was going to take charge of creating business products.

Claude Code, launched last year, is largely responsible for Anthropic’s recent success with businesses. It offers the ability to write and revise code; for some customers, that is a bigger contributor to productivity than the chatbots employees use for more general tasks and search, according to the CEO of a startup that spends millions of dollars per month on both OpenAI and Anthropic. And nonengineers have used the product, as well as its offshoot, Cowork, for noncoding tasks involving other apps they use for work.

The positive reception for Anthropic’s efforts has dialed up the pressure on OpenAI to improve competing products as well as introduce other features prized by large corporate clients. At the same time, some OpenAI products, like agents aimed at knowledge work such as creating and editing spreadsheets, have fallen short of their forecasts.

OpenAI leaders, in meetings with consultants that work with its customers, recently teased upcoming upgrades to OpenAI’s AI coding tool, Codex, which they said would soon beat Claude Code’s performance and features, according to one of the consultants. The company also has been adding collaboration and other workplace features to ChatGPT. That hasn’t gone unnoticed by Microsoft, which is racing to improve its own AI features in its workplace apps.

OpenAI has also been launching features aimed at specific industries such as healthcare. Altman on Friday said in an X post the company was launching features in Codex to focus on areas such as cybersecurity, a field Anthropic has also focused on.

Over the last year, OpenAI has also reorganized the way it sells to customers by having a single salesperson sell multiple products to a customer, versus having multiple people represent various OpenAI products, according to a person with knowledge of the company’s teams.

OpenAI has also been analyzing the areas in which business customers are obtaining the most value from using its AI products, and it is considering releasing tools to help customers better understand the financial benefits they’re getting, said a person familiar with the efforts. Having such tools will be key to its ability to continue landing large agreements with customers, said a consultant who works with OpenAI clients.

The company has struck multiyear deals worth at least $100 million with seven customers, according to the person with knowledge of the matter. Six other customers have inked agreements worth at least $75 million, and these could potentially grow to $100 million or more when the customers renew their contracts, said the person.

Anthropic, in contrast, said in December it had at least nine customers spending more than $100 million annually on its products, including Microsoft, which is on track to spend $500 million annually on Anthropic models.

Anthropic also seems to have pulled ahead of OpenAI in designing contracts for enterprise customers, said the startup CEO and a consultant who works with OpenAI customers. Anthropic lets customers commit to using a set amount of API capacity up front in exchange for discounted pricing, the person said. Some OpenAI customers have privately complained that they haven’t been able to get this level of contracting flexibility.

Last year, though, OpenAI introduced more flexibility. It now provides companies credits for advanced models and tools, which means the per-seat price differs based on the number of credits customers buy, according to the person with knowledge of the company’s enterprise plans. Large enterprise customers commonly get discounts when they strike large commitments, the person said.

Other efforts have stumbled. Last year, the company cut its revenue expectations from selling agents by half, to $1.4 billion in 2025. For instance, ChatGPT Agent, a set of features OpenAI launched in July for ChatGPT subscribers to create and edit spreadsheets and presentations, didn’t meet some internal benchmarks—such as use by 10% of ChatGPT weekly active users, said a person with knowledge of the effort.

Such setbacks haven’t deterred the company, which has long had ambitions to make itself relevant to businesses as well as consumers. Altman has referred to ChatGPT as a “supersmart personal assistant for work.” A year ago, for instance, OpenAI showed off a prototype sales-related AI agent that could sort through sales leads and determine which ones were worth contacting.

OpenAI already generates roughly 40% of its revenue from business customers. Chief Financial Officer Sarah Friar said on Wednesday at Davos that roughly 50% of OpenAI’s business will come from enterprise customers by the end of the year. Altman said in an X post on Thursday that the company had added $1 billion in annualized revenue from API sales in the last month alone. The statistic countered a public perception of the company: “People think of us mostly as ChatGPT,” he noted.