OpenAI unveiled Jalapeño, its first custom AI accelerator, on Wednesday. The chip was designed specifically for large language model inference and is expected to power workloads behind ChatGPT, Codex, the API and future AI products.
• Broadcom shares are showing limited movement. What’s next for AVGO stock?
The announcement immediately sparked discussion about what it means for NVIDIA Corp (NASDAQ:NVDA). But investors may be overlooking another winner hiding in plain sight: Broadcom Inc. (NASDAQ:AVGO).
While OpenAI designed the chip around its own AI workloads, Broadcom helped bring the project to life, providing silicon implementation, networking technologies and production expertise. In fact, OpenAI described Jalapeño as the first accelerator in a multi-generation compute platform being developed alongside Broadcom.
The AI Chip Story Is Getting Bigger Than Nvidia
For much of the AI boom, the playbook was simple. Companies needed more computing power, and Nvidia supplied the hardware. That dynamic is beginning to evolve.
Alphabet Inc. (NASDAQ:GOOGL) (NASDAQ:GOOG) has spent years developing its Tensor Processing Units, or TPUs. Amazon.com, Inc. (NASDAQ:AMZN) built Trainium and Inferentia. Microsoft Corp (NASDAQ:MSFT) introduced Maia. Meta Platforms, Inc. (NASDAQ:META) has been developing its own MTIA accelerators.
Now OpenAI has joined the list.
The shift suggests the industry’s largest AI companies increasingly want hardware optimized for their own models and workloads rather than relying entirely on off-the-shelf chips.
That trend may not be bad news for Nvidia, with its GPUs remaining critical for training frontier models. But it could create a significant opportunity for companies helping customers design and deploy custom silicon.
Why Broadcom Keeps Showing Up In The AI Conversation
Broadcom has quietly emerged as one of the biggest beneficiaries of the custom chip movement.
Unlike Nvidia, which sells its own accelerators, Broadcom’s opportunity lies in helping customers build theirs. The company has become a key partner for hyperscalers and AI companies seeking specialized hardware tailored to specific workloads.
OpenAI’s announcement highlighted Broadcom’s role in networking, chip implementation and large-scale production systems. The release also referenced Broadcom’s Tomahawk networking technology and discussed plans for deployments at gigawatt-scale data centers beginning in 2026.
That matters because AI infrastructure is becoming about more than raw compute power. Networking, memory movement, energy efficiency and system-level optimization are increasingly important as companies look to scale AI services to hundreds of millions of users.
The Bigger Investor Takeaway
The most interesting part of OpenAI’s announcement may not be the chip itself. It may be what the launch says about the direction of the AI industry.
As companies move from training models to serving them at massive scale, custom silicon is becoming a larger part of the conversation. OpenAI’s Jalapeño joins a growing list of purpose-built AI accelerators designed to improve efficiency, reduce costs and optimize performance for specific workloads.
If that trend continues, Broadcom could find itself in an increasingly valuable position.
OpenAI may have launched the chip. But the announcement also reinforced a growing reality for investors: the more AI companies build their own silicon, the more important Broadcom’s role in the ecosystem could become.
Photo: Tada Images / Shutterstock
Login to comment