OpenAI, one of the most notable companies in the field of artificial intelligence in recent years, manages to stay on the agenda not only with its software development but also with the huge infrastructures it has established. Multi-layered collaborations with Oracle, Nvidia and AMD have significantly increased the company’s growth rate. However, it became clear with the latest statements of CEO Sam Altman that these developments were only the beginning. Altman raised expectations by stating that much larger agreements will be announced in the coming period.
Nvidia CEO Jensen Huang detailed these developments in a program he attended on CNBC. Especially OpenAI’s extraordinary collaboration with AMD was also on Huang’s agenda. Drawing attention with the statement “I was not fully aware of the agreement”, Huang pointed out that this new structure could change the balance in the sector. However, Huang spoke not only of competition but also of changing strategic balances. At this point, it is possible to see that interactions between companies are no longer solely focused on competition, but are shaped through the investment model.
OpenAI plans new hardware investments to build independent data centers
The most striking aspect of the agreement between OpenAI and AMD was that the investment model went beyond the ordinary. Thanks to this structure, OpenAI will be able to own up to 10% of AMD shares as long as certain conditions are met. In return, contribution will be made to the development process of AMD’s new generation artificial intelligence GPUs. In other words, OpenAI becomes not only a technology buyer, but also a technology partner involved in the process. Thanks to this cooperation, both parties aim for long-term strategic gains.
The cooperation with Nvidia has a completely different structure. This time, Nvidia invested in OpenAI and took a direct stake in the company. Thus, a company in the position of a manufacturer secured the future of its own products by providing capital support to a company in the position of a user. This investment approach of Nvidia has transformed the classical supply relationship. But what is remarkable is that these relations are beginning to deepen technically as well as financially.
Until now, OpenAI has been using Nvidia hardware through cloud providers such as Microsoft Azure, Oracle OCI and CoreWeave. Despite this, it was stated that there will now be a transition to the direct supply model. Nvidia is not just GPU; It will provide all of the system infrastructure, network equipment and artificial intelligence-specific components to OpenAI. This transition directly supports OpenAI’s ambition to have its own data centers in the medium term. In other words, a plan to use own resources instead of outsourced services is being put into effect.
Jensen Huang also touched upon the high cost of these investments. According to him, the cost of installing a gigawatt AI data center varies between $50 and $60 billion. This figure includes not only hardware but also infrastructure, energy and land costs. Despite this, OpenAI started the construction of a total of 10 gigawatt data centers in the USA in 2025. These data centers are being implemented within the scope of the 500 billion dollar Stargate agreement with Oracle and SoftBank.
In addition to this project, the $300 billion cloud agreement signed with Oracle also draws attention. On the other hand, an infrastructure investment of 10 gigawatts was planned within the scope of the agreement with Nvidia. The collaboration with AMD reaches a capacity of 6 gigawatts. In light of all this data, the total value of OpenAI’s infrastructure investments in 2025 has approached 1 trillion dollars. The Stargate UK initiative in Europe and the United Kingdom has also been included in this expansion.
However, some circles find these investment models “circular” and lacking in transparency. Especially in the agreements made with Nvidia and AMD, comments that the companies bear each other’s financial burdens come to the fore. These criticisms voiced by Bloomberg suggest that the investment structure may lead to conflicts of interest over time. Still, despite these comments, it is noteworthy that companies continue to increase their cooperation. This shows that not only competition but also interdependencies are at the forefront in the sector.
Sam Altman detailed these collaborations in the podcast published by Andreessen Horowitz. According to Altman, the new models OpenAI is working on will be much more powerful and comprehensive. This means that the need for infrastructure will increase exponentially. Company revenues are not yet at this level; It is stated that an income of 4.5 billion dollars was achieved in the first half of 2025. However, Altman believes that these investments will bring great economic returns in the long run.