OpenAI expected to burn $115B on ChatGPT push by 2029

OpenAI has elevated its cash burn forecast this year through 2029 to a total of $115 billion. The company’s recent cash burn expectation is also $80 billion higher than it previously projected. 

According to a report by The Information, the surge in cash burn for OpenAI comes at a time when it’s ramping up spending to power the artificial intelligence behind its popular ChatGPT chatbot. The tech firm has also become one of the world’s biggest renters of cloud services.

OpenAI plans to develop its chips and data center facilities

The source revealed that the AI company expects to burn over $8 billion this year. OpenAI had forecasted early in the year that it would only burn around $1.5 billion.

According to the report, OpenAI doubled its cash burn expectations for 2026 to more than $17 billion, surpassing its previous forecast of $10 billion. The firm also projects a $35 billion cash burn in 2027 and $45 billion in 2028. 

The FT also disclosed on Thursday that the Silicon Valley startup plans to develop its data center server chips and facilities to power its technology. According to the report, the initiative aims to control the tech company’s surging operational costs.

The firm relies on substantial computing power to train and run its systems. The company’s CEO, Sam Altman, has also advocated the need for increased computing power to accommodate the growing demand for AI products such as ChatGPT.

Deloitte’s 2025 AI infrastructure Survey revealed that the energy demands of AI are straining traditional power grids. According to the study, 79% of executives anticipate increased power demand through the next decade, with grid stress emerging as a top challenge.

The source added that U.S. semiconductor giant Broadcom will partner with OpenAI to produce the first set of chips and start shipping them by next year. Also, OpenAI allegedly plans to use the chips internally rather than selling them for external clients. 

Broadcom’s CEO, Hock Tan, hinted the company had partnered with an undisclosed customer that committed to $10 billion in orders. During a call with analysts, he revealed the firm had secured a fourth customer to boost its custom AI chip division. Tan stated the collaboration with OpenAI has enhanced its growth outlook for fiscal 2026 by generating immediate and substantial demand. 

OpenAI partners with Broadcom to produce chips

OpenAI also partnered with Broadcom and Taiwan Semiconductor Manufacturing Co. (TSMC) nearly a year ago to develop its first in-house chip. The firm was also planning to add AMD chips alongside Nvidia chips to meet its surging infrastructure demands.

OpenAI revealed in February plans to reduce its reliance on Nvidia’s chips. The firm said it will finalize the design of the new chip in the next few months and then send it to TSMC for fabrication. OpenAI’s initiative also builds on its ambitious plans to increase its semiconductor production at the Taiwanese company next year.

According to the report, OpenAI hopes to use the new chips to strengthen its negotiating leverage with other chip suppliers, including Nvidia. The company’s in-house team, led by Richard Ho, will design the chip to produce advanced processors with broader capabilities with each new iteration.

OpenAI collaborated with Oracle in July to launch a 4.5-gigawatt data center. The initiative also complements the firm’s $500 billion Stargate project, including investments from Japanese firm SoftBank Group. The tech giant has also collaborated with Google Cloud to supply computing capacity.

Want your project in front of crypto’s top minds? Feature it in our next industry report, where data meets impact.