- Openai (probably) joins Google, Meta and Bytedance in Broadcom’s custom ASIC partnership
- Broadcom secures $ 10 billion AI -Rack orders as Nvidia is facing new rivals
- Nvidia’s largest customers pursue internal chips with Broadcom that controls the transition
As we have reported more than a new time in the past, the AI hardware market is changing, with some of Nvidia’s largest customers looking for ways to reduce costs and gain more control over their systems.
Instead of only relying on Nvidia’s expensive GPUs, companies begin to design their own ASICs tailored to their workload.
Broadcom is one of the larger players in this room, offering the expertise needed to transform these custom designs into production -ready chips and systems.
Reporting on Broadcom’s financial results for its third quarter, The next platform Says the Silicon supplier has now secured a fourth customer for his custom XPU program, adding to partnerships with Google, Meta and Bytedance.
Industrial reports and timing suggest that this newest client is Openai, who develops his own inference processor known as Titan led by Richard Ho, a former Google TPU engineer.
Nvidia still still dominates the market of course – somehow – with its Blackwell GB300 NVL72, but implementing such rackscale systems is expensive and companies with massive AI models want hardware designed to better match their needs.
Custom ASICs are seen as a way to empty costs while offering greater flexibility than a GPU shelves, and Broadcom is well placed to guide complex accelerator projects through design, production and packaging.
On a call with Wall Street analysts extended CEO Hock Tan on Broadcom’s named fourth client (coughAt OpenaiAt cough), and said, “Now to these three customers, as we had mentioned earlier, we have worked with other prospects of their own AI accelerators.”
“The last quarter released one of these views of production orders to Broadcom, and we have consequently characterized them as a qualified customer for XPUs and have actually secured over $ 10 billion orders from AI stands based on our XPUs,” Tan continued.
“And reflects this, we are now expecting the prospects of our accounting executions from 2026 to improve significantly from what we had indicated last quarter.”
This figure of $ 10 billion refers to complete AI rack systems, not Broadcom’s share of the underlying chip design.
Revenue from these orders is scheduled to begin in the third quarter of the financial year 2026.
It is clear that NVIDIA’s largest buyers are no longer satisfied that only on GPUs, and by investing in ASICs, they bet that custom hardware will bring efficiency and control. With its expertise, Broadcom plays itself as the company that can make these designs come true.
