OpenAI contracted Broadcom in a $10 billion agreement to begin mass production of custom AI chips in 2026. This strategic move aims to control internal compute infrastructure and reduce dependence on suppliers like Nvidia. TSMC will manufacture the new chip, which OpenAI will initially deploy within its internal systems to power training and daily operations, targeting cost reduction. Following the announcement, Broadcom’s shares traded more than 10% higher in premarket trading.
“Ours will be special, and what’s more, it will work,” said an unnamed person not authorized to speak for the company. “All those other 100 or more processors are junk.”
OpenAI has signed a $10 billion chip design deal with Broadcom to start mass production of custom AI chips in 2026. This strategic move marks a shift away from Nvidia and over 100 other AI processor providers, focusing instead on building internal computing infrastructure.
Broadcom shares have increased by nearly 100% over the past year as the chipmaker positions itself within the artificial intelligence sector. The company develops custom chips for Google and other cloud providers. Broadcom shares traded more than 10% higher in premarket trading following the announcement.
OpenAI will launch its first proprietary AI chip in 2026. The project represents a development for the ChatGPT maker as it seeks to control hardware and reduce dependence on external chip suppliers like Nvidia.
Industry sources indicate that the new AI chip will not reach external buyers initially. OpenAI will deploy it within internal systems to power training and daily operations. This approach targets cost reduction and management of increasing computing power demand.
Broadcom’s chief executive, Hock Tan, indicated this partnership when he disclosed that his company had secured over $10 billion in new AI orders from a client. While he did not identify the buyer, industry observers believe it represents OpenAI. Tan projects AI-related sales will grow in 2026 as a result.
This strategy aligns OpenAI with other tech companies like Amazon, Google, and Meta. These firms develop custom chips to manage artificial intelligence workloads more effectively. Creating custom chips allows them to lower costs, boost performance, and tailor hardware to specific needs.
Taiwan Semiconductor Manufacturing Company (TSMC), the world’s contract chipmaker, will manufacture the new AI chip. According to reports, OpenAI has finalized the design and prepared it for production. Alongside this effort, OpenAI continues to utilize AMD and Nvidia chips while building a diverse supply chain.
High-performance processors are in increasing demand, driven by generative AI applications like ChatGPT. While Nvidia continues to dominate with its GPUs in this area, OpenAI’s move marks a new direction for hardware-independent models. Building its own AI chip allows it to control performance, pricing, and supply stability.
Chip development faces challenges. It demands investment, lengthy timelines, and technical risks. Other companies have struggled to meet speed or efficiency goals when designing silicon. This project shows OpenAI’s commitment to preparing for the future with tools it can control.
By 2026, when its first AI chip arrives, the AI company might impact how systems operate and improve cost efficiency to compete more directly with existing participants in the technology sector.
Broadcom shares have increased by nearly 100% over the past year as the chipmaker positions itself within the artificial intelligence sector.
If you want to learn everything about the over 100 AI processor and IP suppliers, click on this link to get our quarterly report, which includes a detailed database of who’s who, what they do, and how they do it.
WHAT DO YOU THINK? GOOD ENOUGH TO TELL YOUR FRIENDS ABOUT IT? YOU HAVE OUR PERMISSION TO SHARE IT.