OpenAI–Broadcom alliance signals a shift to open infrastructure for AI

OpenAI has partnered with Broadcom to co-develop and deploy its first in-house AI processors. The move could reshape data center networking dynamics and chip supply strategies as the ChatGPT maker races to secure more computing power for its rapidly growing AI workloads.

The multi-year collaboration will deploy 10 gigawatts of OpenAI-designed accelerators and Broadcom’s Ethernet-based networking systems starting in 2026, underscoring a move toward custom silicon and open networking architectures that could influence how enterprises build and scale future AI data centers.

“By designing its own chips and systems, OpenAI can embed what it’s learned from developing frontier models and products directly into the hardware, unlocking new levels of capability and intelligence,” the two companies said in a statement. “The racks, scaled entirely with Ethernet and other connectivity solutions from Broadcom, will meet surging global demand for AI, with deployments across OpenAI’s facilities and partner data centers.”

Ethernet’s AI advantage grows

The decision to rely on Broadcom’s Ethernet fabric, rather than Nvidia’s InfiniBand interconnects, signals OpenAI’s intent to build a more open and scalable networking backbone that could set a precedent for AI infrastructure across hyperscale and enterprise environments.

Analysts suggest that this is in line with a broader industry momentum toward open networking standards, which can deliver flexibility and interoperability.

“OpenAI’s choice signals a shift toward more open, cost-efficient, and scalable architectures,” said Charlie Dai, VP and principal analyst at Forrester. “Ethernet offers broader interoperability and avoids vendor lock-in, which could accelerate the adoption of disaggregated AI clusters. This move is another attempt to challenge InfiniBand’s dominance in high-performance AI workloads and may push hyperscalers to standardize on Ethernet for ecosystem diversity and digital sovereignty.”

The decision also reflects a future of AI workloads running on heterogeneous computing and networking infrastructure, said Lian Jye Su, chief analyst at Omdia.

“While it makes sense for enterprises to first rely on Nvidia’s full stack solution to roll out AI, they will generally integrate alternative solutions such as AMD and self-developed chips for cost efficiency, supply chain diversity, and chip availability,” Su said. “This means data center networking vendors will need to consider interoperability and open standards as ways to address the diversification of AI chip architecture.”

Hyperscalers and enterprise CIOs are increasingly focused on how to efficiently scale up or scale out AI servers as workloads expand. Nvidia’s GPUs still underpin most large-scale AI training, but companies are looking for ways to integrate them with other accelerators.

Neil Shah, VP for research at Counterpoint Research, said that Nvidia’s recent decision to open its NVLink interconnect to ecosystem players earlier this year gives hyperscalers more flexibility to pair Nvidia GPUs with custom accelerators from vendors such as Broadcom or Marvell.

“While this reduces the dependence on Nvidia for a complete solution, it actually increases the total addressable market for Nvidia to be the most preferred solution to be tightly paired with the hyperscaler’s custom compute,” Shah said.

Most hyperscalers have moved toward custom compute architectures to diversify beyond x86-based Intel or AMD processors, Shah added. Many are exploring Arm or RISC-V designs that can be tailored to specific workloads for greater power efficiency and lower infrastructure costs.

Shifting AI infrastructure strategies

The collaboration also highlights how networking choices are becoming as strategic as chip design itself, suggesting a change in how AI workloads are powered and connected.

OpenAI’s move underscores a broader industry shift toward diversifying supply chains and ensuring better control over performance and cost.

“This partnership underscores a growing trend to reduce dependency on Nvidia’s GPUs and proprietary stack,” Dai added. “As AI adoption continues to scale and AI leaders seek more balance between performance gains and cost control, vertical integration through custom silicon becomes strategic. This could elevate ASICs and Ethernet-based fabrics and foster competition among chipmakers.” However, Su noted that only a handful of enterprises, mainly hyperscalers and large GenAI vendors, will be able to design their own AI hardware while providing sufficient internal software support. Most enterprises will likely continue to rely on Nvidia’s full-stack solutions.