Key Takeaways:

• OpenAI plans to begin mass production of its custom AI chips with Broadcom in 2026.
• The chips will be used internally to support OpenAI’s training and inference workloads, not sold to outside customers.
• The move is designed to reduce reliance on Nvidia and optimize performance for OpenAI’s models.
• Google, Amazon, and Meta have already invested in developing proprietary chips, setting a precedent.
• OpenAI’s strategy raises questions about whether AI companies will follow a path of vertical integration similar to Apple.

OpenAI is preparing to manufacture its own AI chips in partnership with Broadcom beginning in 2026, according to a report from the Financial Times and confirmed by Reuters. The move comes as demand for computational power continues to outpace available supply, with Nvidia dominating the market for high-performance GPUs. By developing custom silicon, OpenAI is signaling an intent to gain greater control over its infrastructure and reduce its reliance on external suppliers.

Why OpenAI is Building Chips

Reports earlier this year indicated that OpenAI finalized the design of its first in-house processor and submitted it to TSMC for fabrication. Mass production is now scheduled for 2026. These chips are not intended for sale to other companies but will instead be deployed within OpenAI’s own data centers to support training and inference for advanced language models.

There are several reasons driving the shift. First, Nvidia’s GPUs remain both expensive and supply-constrained, creating challenges for scaling. By working with Broadcom, OpenAI could gain more predictable access to hardware tailored specifically for its workloads. Second, designing chips in-house gives OpenAI the ability to optimize performance per watt, an increasingly important factor given the enormous energy costs associated with training frontier models. Third, control over hardware allows OpenAI to pursue long-term strategies without being as vulnerable to supply chain bottlenecks or competitive pricing pressures from a single dominant supplier.

A person familiar with the project told the Financial Times that “OpenAI wants to ensure its long-term ability to scale. The cost of relying solely on Nvidia is simply too high.”

Lessons From Other Big Tech Players

OpenAI is not alone in building its own processors. Google has long relied on its Tensor Processing Units, which are optimized for AI training and inference. Amazon developed the Graviton and Trainium processors for AWS, giving customers access to compute that is customized for cloud workloads. Meta has invested in its own accelerators to support its AI initiatives. Each of these companies has moved toward designing chips to better align hardware with their specific use cases.

The common theme is that AI workloads demand such specialized performance that off-the-shelf GPUs are no longer sufficient on their own. Custom chips allow companies to reduce costs, improve efficiency, and retain tighter control of their technological roadmap. OpenAI’s collaboration with Broadcom fits squarely within this trend.

The Apple Comparison

The news raises a larger question: are AI companies on a path toward vertical integration similar to Apple’s model? Apple has long controlled its supply chain, from designing custom processors to overseeing software, hardware, and even distribution. This end-to-end control allows Apple to deliver integrated products while capturing a larger share of value.

For OpenAI, the comparison is more limited. The company is not building consumer hardware, nor is it aiming to sell chips to other firms. Instead, the vertical integration is focused on the back end—developing silicon optimized for training and running its AI models. In that sense, OpenAI’s strategy is closer to Google or Amazon than Apple. The goal is not to create an ecosystem of consumer devices but to secure the compute resources necessary to operate increasingly large and complex AI systems.

As one industry analyst noted, “This is not about copying Apple’s consumer model. It’s about survival in a world where demand for compute will keep doubling every year. If you don’t own some part of your stack, you risk being locked out.”

Implications for the AI Landscape

OpenAI’s move could reshape dynamics in the AI industry in several ways. First, it strengthens Broadcom’s position as a key supplier. Broadcom has already seen its AI chip business expand rapidly, with more than $10 billion in orders attributed to new partnerships. While Nvidia remains the market leader, Broadcom’s work with OpenAI signals growing competition in the space.

Second, the shift underscores the importance of hardware sovereignty for AI firms. As generative models increase in scale, owning the compute layer may become as important as developing the software itself. This could pressure smaller AI startups, which lack the resources to build custom chips, to form alliances or rely more heavily on cloud providers that offer proprietary hardware.

Third, OpenAI’s strategy highlights the capital intensity of competing at the frontier of AI. Designing chips requires significant upfront investment, long lead times, and specialized expertise. The fact that OpenAI is committing to this path suggests confidence in the sustained demand for large-scale AI systems.

What Comes Next

The first chips are expected to roll off production lines in 2026, though the full impact may take years to unfold. OpenAI will need to integrate these processors into its existing infrastructure and prove that they deliver meaningful advantages over Nvidia’s offerings. If successful, the company could reduce its dependence on external suppliers while improving cost efficiency.

Whether this represents the beginning of a broader wave of vertical integration among AI firms remains to be seen. Some may follow Google, Amazon, Meta, and now OpenAI in pursuing custom silicon. Others may prefer to rely on partnerships with established chipmakers, particularly if their focus is on software applications rather than foundational models.

What is clear is that hardware has become inseparable from the future of AI. As workloads grow and competition intensifies, companies at the forefront will likely look for ways to control as much of the stack as possible. OpenAI’s decision to move into chip design and production is not a bid to become Apple, but it does signal a deeper convergence of hardware and software in the AI era.

Learn how AI Agents can supercharge your company’s profits and productivity at TMC’s AI Agent Event Sept 29-30, 2025 in DC.

Rich Tehrani serves as CEO of TMC and chairman of ITEXPO #TECHSUPERSHOW Feb 10-12, 2026 and is CEO of RT Advisors and is a Registered Representative (investment banker) with and offering securities through Four Points Capital Partners LLC (Four Points) (Member FINRA/SIPC). He handles capital/debt raises as well as M&A. RT Advisors is not owned by Four Points.

The above is not an endorsement or recommendation to buy/sell any security or sector mentioned. No companies mentioned above are current or past clients of RT Advisors.

The views and opinions expressed above are those of the participants. While believed to be reliable, the information has not been independently verified for accuracy. Any broad, general statements made herein are provided for context only and should not be construed as exhaustive or universally applicable.

Portions of this article may have been developed with the assistance of artificial intelligence, which may have contributed to ideation, content generation, factual review, or editing.