OpenAI is deepening its relationship with Samsung Electronics as it develops its own next-generation AI chips, expanding a partnership that already spans advanced memory, AI infrastructure and one of OpenAI’s largest enterprise deployments.
The latest development was disclosed on September 9 by Harrison Kim, General Manager of OpenAI Korea, who said the companies have made significant progress in joint work around next-generation chips. Samsung did not provide additional customer-specific details.
What happened?
Speaking in Seoul, Kim said joint production and research on the next-generation chips OpenAI is developing is one of the areas where cooperation with Samsung Electronics has progressed most strongly. The companies have not yet disclosed the precise technical scope of Samsung’s role in those chips.
That distinction matters. OpenAI has separately announced its custom AI chip work with Broadcom, with TSMC manufacturing the chips. The latest comments therefore should not be read as confirmation that Samsung has replaced TSMC as the manufacturer of OpenAI’s processor. Instead, they point to a broader and deepening semiconductor relationship whose exact responsibilities have not yet been publicly detailed.
Samsung was already a major part of OpenAI’s infrastructure plans
This relationship did not start today. In October 2025, OpenAI announced Samsung and SK as strategic partners in its Stargate AI infrastructure initiative. Samsung Electronics agreed to work as a strategic memory partner supplying advanced semiconductor solutions for OpenAI’s infrastructure.
OpenAI said at the time that projected memory demand for Stargate could reach as much as 900,000 DRAM wafer starts per month as infrastructure scales. Samsung’s capabilities across high-performance memory, logic, foundry services, advanced packaging and heterogeneous integration make it strategically important to the build-out of large AI systems.
Why AI memory matters almost as much as the processor
Discussion of AI hardware often focuses on processors from companies such as NVIDIA. But advanced memory is another critical part of the system. Large AI workloads need enormous quantities of data to move rapidly between memory and processors, making high-bandwidth and energy-efficient memory increasingly valuable.
Kim said demand for memory chips is expected to keep increasing as more advanced chips are needed for faster and more complex computing. That makes relationships with major South Korean semiconductor suppliers strategically important to OpenAI.
Samsung is also one of OpenAI’s biggest ChatGPT customers
The relationship extends well beyond hardware. OpenAI announced in June that Samsung Electronics was rolling out ChatGPT Enterprise and Codex to all Samsung Electronics employees in Korea and to Device eXperience employees worldwide, describing it as one of OpenAI’s largest enterprise deployments to date.
Samsung is using the tools across areas including research and development, software development, manufacturing, product development, marketing and corporate functions. That means the two companies now have a relationship running in both directions: Samsung can provide technology needed to build AI infrastructure, while OpenAI provides AI software used inside Samsung itself.
Why this matters for the AI chip race
AI companies increasingly want greater control over the hardware beneath their models. Custom silicon can be optimised for particular workloads and may reduce reliance on a single external accelerator supplier as computing demand grows.
OpenAI’s custom-chip programme is therefore about more than building another processor. At scale, the company needs a supply chain spanning chip design, fabrication, memory, packaging, networking, power and data centres. Samsung is one of the few technology groups with expertise across several of those layers.
What does it mean for ChatGPT users?
Consumers should not expect an immediate change to ChatGPT because of today’s comments. The impact is longer term.
If OpenAI can build more efficient AI infrastructure and secure enough advanced chips and memory to support demand, it could ultimately help the company serve increasingly capable models at larger scale. Hardware efficiency also influences how quickly AI services respond, how much they cost to operate and how widely advanced features can be deployed.
Samsung’s AI semiconductor push is getting broader
The OpenAI development arrives as Samsung is accelerating AI adoption inside its own semiconductor operations. Samsung also announced this week a strategic partnership with Mistral AI to deploy customised on-premises AI across semiconductor engineering and manufacturing.
Taken together, the announcements show Samsung positioning itself on both sides of the AI boom: as a supplier of the physical semiconductor infrastructure needed to run advanced AI and as a major user of AI to improve how those semiconductors are designed and manufactured.
The bigger picture
The competition between OpenAI, Google, Meta, Amazon and other major AI players is increasingly becoming a competition over infrastructure as well as models. Securing processors is only part of that challenge. Memory supply, packaging capacity, fabrication and energy are becoming equally strategic.
OpenAI and Samsung already had substantial ties. The important development now is that OpenAI says cooperation is progressing around the next generation of chips it is building. Until either company discloses the technical details, Samsung’s exact role should not be overstated — but the direction is clear: the AI software race and semiconductor race are becoming increasingly intertwined.
Sources: OpenAI, Samsung Electronics and Reuters. The Tech Boom has distinguished confirmed partnership details from areas where the companies have not yet disclosed technical specifics.
