OpenAI is taking a more direct role in the hardware behind its AI systems. According to Reuters, the company is developing its first proprietary chip in collaboration with Broadcom and TSMC, while also expanding its use of AMD processors alongside Nvidia chips.
The move points to a broader effort to manage the hardware pressure that comes with large-scale AI. OpenAI has looked at several ways to diversify its chip supply and reduce costs, but for now it is focusing on chip development rather than building its own network of chip factories.
Why OpenAI is building its own AI hardware
AI systems depend on large amounts of computing infrastructure. As demand grows, the chips that power those systems become a central constraint. The source article says OpenAI is responding to rapidly growing infrastructure demand by widening its chip options and working on proprietary hardware.
That does not mean OpenAI is walking away from existing chip suppliers. The article says AMD processors are being added alongside Nvidia chips. In practical terms, that suggests a strategy built around more than one hardware path.
The proprietary chip effort is another part of the same picture. A custom chip can give a company more control over the hardware it uses for its own AI systems. The source does not provide technical details about the chip, but it does identify the project as OpenAI's first proprietary chip.
Broadcom and TSMC shape the plan
OpenAI is not developing the chip in isolation. Broadcom is helping with chip development, according to the source article. Broadcom is also helping OpenAI secure manufacturing capacity at TSMC.
That division of work matters. OpenAI is focusing on the chip it wants for its AI systems, while working with established partners for development support and manufacturing access. The article does not describe the chip's design, performance targets, or timeline, so the clearest confirmed point is the structure of the collaboration: OpenAI, Broadcom, and TSMC are each part of the effort.
TSMC's role is described specifically as manufacturing capacity. The source does not say OpenAI will own factories or directly manufacture the hardware itself. Instead, the plan centers on developing its own chips while using external manufacturing support.
Why chip factories are off the table for now
OpenAI considered a more ambitious route: building its own network of chip factories. The source says that option has been ruled out for the time being because of the high cost and the time involved.
That decision narrows the company's hardware strategy. Rather than taking on factory construction and operation, OpenAI is prioritizing proprietary chip development and partnerships. It is a more focused approach to the same problem: securing enough computing power while trying to diversify supply and reduce costs.
The difference is important. Building a network of factories would be a much broader infrastructure commitment. Developing chips with partners keeps attention on the hardware design and manufacturing access that OpenAI needs for its AI systems.
AMD joins Nvidia in OpenAI's chip mix
The source article also says OpenAI is expanding its AI chip offerings with AMD processors to meet rapidly growing infrastructure demand alongside Nvidia chips. That puts AMD into the company's broader hardware mix rather than replacing Nvidia outright.
For OpenAI, using more than one processor source can support diversification. The source directly ties this to the company's interest in diversifying chip supply and reducing costs. It does not provide purchasing details, deployment numbers, or product names, so the confirmed point is strategic: OpenAI is widening the set of chips available for its AI infrastructure.
This matters because AI hardware is not just a background concern. It affects how companies plan capacity, manage costs, and support future systems. The article frames OpenAI's hardware decisions as a response to infrastructure demand that is rising quickly.
What this signals for OpenAI's infrastructure strategy
OpenAI's current direction has three parts: develop a proprietary AI chip, work with Broadcom and TSMC, and add AMD processors alongside Nvidia chips. Taken together, those moves show a company trying to gain more flexibility in the hardware layer that supports its AI systems.
The plan is also selective. OpenAI considered building its own chip factory network, but that route is not moving forward for now. Instead, it is leaning on partners while pursuing a chip of its own.
Based on the source article, the key takeaway is not that OpenAI is abandoning outside suppliers. It is that OpenAI is building a more diversified hardware strategy. Proprietary chip development, external manufacturing capacity, and a broader processor mix all point toward the same goal: meeting infrastructure demand while reducing dependence on a single path for AI chips.