#ai-accelerators

#ai-accelerators

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 racks will be "scaled entirely with Ethernet and other connectivity solutions from Broadcom... with deployments across OpenAI's facilities and partner datacenters."

Just days after announcing a $1.1 billion Series G funding round, AI chip startup Cerebras Systems pulled its S-1 IPO filing without so much as an explanation. Amid growing concerns of an AI bubble, has Cerebras CEO Andrew Feldman gotten cold feet? Apparently not. Taking to LinkedIn on Sunday, Feldman apologized for the lack of transparency, and presumably any panic the decision might have caused. "On Friday, Cerebras Systems withdrew our S-1. We didn't explain why - that was a mistake," he wrote, adding that he still has every intention of taking the company public.

First off the rank, in the first quarter of 2026, will be the Ascend 950PR which, according to slideware shown at the conference, will boast one petaflop performance with the 8-bit floating-point (FP8) computation units used for many AI inferencing workloads. The chip will also include 2 TB/s interconnect bandwidth and 128GB of 1.6 TB/s memory. In 2026's final quarter Huawei plans to deliver the 950DT, which will be capable of two petaflops of FP4 performance thanks to the inclusion of 144GB of 4 TB/s memory.

"We must do better at maintaining and expanding our position in the global market, while safeguarding America's technological edge. With these enhanced security measures, we can continue to expand access to US technology without compromising our national security."

AMD's recent $800 million charge highlights the impact of U.S. trade restrictions impacting its Instinct MI308 AI accelerators, demonstrating the evolving landscape of the semiconductor industry.