OpenAI has significantly increased its projected cash burn to $115 billion through 2029 as it scales up investments to power the artificial intelligence behind its widely used ChatGPT chatbot, according to a report on Friday.
The revised estimate is around $80 billion higher than earlier projections. The company, one of the largest renters of cloud servers globally, now expects to spend over $8 billion in 2025—about $1.5 billion more than its previous forecast earlier this year, the report added.
To rein in mounting expenses, OpenAI is working on developing its own server chips and data center facilities, according to The Information. The company will unveil its first AI chip next year in collaboration with U.S. semiconductor major Broadcom, with plans to use the chip internally rather than sell it commercially, the Financial Times reported.
In July, OpenAI expanded its partnership with Oracle to add 4.5 gigawatts of data center capacity as part of its larger Stargate initiative—an ambitious project worth up to $500 billion and targeting 10 gigawatts, backed by SoftBank.
The company has also brought Google Cloud on board as a computing capacity provider. The Information reported that OpenAI expects its cash burn to more than double to over $17 billion in 2026—$10 billion higher than its earlier forecast—and to rise further to $35 billion in 2027 and $45 billion in 2028.
