OpenAI plans to invest $100 billion in building backup servers

OpenAI is planning a massive infrastructure investment program, expected to spend approximately $100 billion over the next few years on building and leasing backup servers to cope with the rapid growth in demand for AI computing power and potential server supply disruptions.

This $100 billion investment, separate from existing computing power and data center leasing costs, will reportedly be used primarily to build more robust redundancy systems to improve service stability and mitigate delays in new feature releases caused by insufficient computing power.

Due to surging customer demand, OpenAI has repeatedly delayed the release of new features or models due to insufficient server capacity. Furthermore, this investment may be related to Project Stargate, a US-based AI infrastructure hyperscale initiative jointly launched by OpenAI, Oracle, SoftBank, and others.

With an estimated total investment of $500 billion, Stargate aims to build multiple high-capacity computing centers to support large-scale AI model training. As a key component of this initiative, the investment in backup servers is expected to further enhance the resilience and reliability of the overall infrastructure.

Once launched, the project will gradually deploy backup servers over the next five years to alleviate computing power constraints in cloud services and model training, while also addressing potential disruptions such as supply chain fluctuations, lease contract changes, geopolitical risks, and hardware failures.