OpenAI is building software to break free from Nvidia’s grip on AI infrastructure

The AI giant is developing tools to run workloads across chips from multiple providers, partnering with AMD and Broadcom while designing its own custom silicon.

Share

Add us on Google by Editorial Team Jun. 1, 2026

OpenAI is building internal software that lets its AI workloads run across hardware from multiple manufacturers, a direct challenge to Nvidia’s near-monopoly on the infrastructure powering modern artificial intelligence.

The effort centers on creating low-level platform primitives, essentially the foundational code that tells different chips how to handle the same AI tasks.

The multi-vendor playbook

This isn’t just a software project. OpenAI is pairing its internal tooling with major hardware partnerships designed to give it real alternatives to Nvidia’s ecosystem.

Advertisement

In October 2025, OpenAI formalized a multi-year partnership with Broadcom focused on co-developing accelerator and networking systems. That same month, the company signed a deal with AMD to deploy up to 6 gigawatts of AMD Instinct GPUs, with deployment starting in the second half of 2026.

The company’s Hardware organization is also designing its own silicon solutions, working with TSMC on chiplet architectures and advanced packaging technologies. Custom inference chips are slated for production by 2026, giving OpenAI yet another option beyond buying off the shelf from Nvidia.

Why this matters beyond OpenAI

OpenAI isn’t the first company to pursue hardware diversification. Google has its TPU chips. Amazon has Trainium and Inferentia. Microsoft has its own custom AI accelerator, Maia.

The Stargate initiative, OpenAI’s massive data center scaling effort, provides the physical backdrop for all of this.

The job postings tell the story. OpenAI’s Compute Infrastructure team is explicitly hiring for roles focused on building automatable, repeatable systems for heterogeneous clusters.

What this means for investors

AMD’s deal for 6 gigawatts of Instinct GPU deployment is a massive vote of confidence in its AI hardware roadmap. Broadcom’s multi-year co-development partnership positions it as a key player in the networking and accelerator layers that connect everything together.

OpenAI designing its own inference chips with TSMC mirrors what Google did with TPUs years ago. If OpenAI achieves similar results by 2026, it would reduce its per-query costs substantially, potentially changing the economics of offering AI services at consumer price points.

Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.