OpenAI’s advertising business, still in its early stages, has surpassed $100 million in annual recurring revenue less than two months into its U.S. market pilot. In January, the artificial intelligence startup announced plans to test advertising with free users and ChatGPT Go subscribers in the U.S., a move that could open up a significant new revenue stream. Rival Anthropic seized the opportunity to mock the effort, making OpenAI’s ad push the focal point of its first Super Bowl marketing campaign.
According to an OpenAI spokesperson, the company is currently working with more than 600 advertisers, and privacy-related trust metrics have remained unaffected. Meanwhile, OpenAI has begun exploring additional tests in Canada, Australia and New Zealand. Per the announcement made in January, ads within ChatGPT will appear at the bottom of the chatbot’s responses, will not influence the content of the answers, and will be clearly labeled. Users under 18 will not see ads, nor will ads appear alongside specific topics such as politics, health and mental health. The spokesperson said Thursday that in the U.S. market, roughly 85 percent of free and Go users are eligible to see ads, but the actual daily display rate is below 20 percent.
Although the pilot has shown early success, some advertisers have expressed frustration over the slow, cautious pace of rollout. OpenAI said the gradual approach to the ad initiative is intentional, with the current focus on learning and optimizing the user experience before pursuing broader expansion.
Over the past few years, Nvidia has dominated the AI hardware market, with its graphics processors helping make it the world’s most valuable company. But the landscape is shifting. A number of AI software developers are accelerating efforts to develop their own hardware in an attempt to reduce dependence on Nvidia. Google parent company Alphabet is a prime example—its tensor processing units compete directly with Nvidia’s GPUs in certain AI applications. OpenAI, the company behind ChatGPT, is also seeking alternatives and has turned its attention to Broadcom (AVGO).
Broadcom specializes in collaborating with companies to develop custom chips and has served as a partner to Google in developing TPUs, while also working with tech giants such as Microsoft, Amazon and Meta. Late last year, OpenAI entered into a multiyear agreement with Broadcom to jointly develop a 10-gigawatt custom AI accelerator—a type of chip better tailored to the specific needs of OpenAI’s software compared with Nvidia’s general-purpose hardware. The move not only puts pressure on Nvidia but also reflects a broader industry shift from GPUs to custom chips. OpenAI’s main rival, Anthropic, is pursuing a similar strategy, expanding its use of Google Cloud while incorporating computing power based on TPU chips developed through the partnership between Google and Broadcom.
The partnership with OpenAI carries significant weight for Broadcom and positions the company as a key player to watch in the AI chip sector.
