Huawei Technologies Co. openly admits its silicon can’t match Nvidia Corp.’s in raw power and speed.

In theory, the technique lets Huawei link as many as 15,488 of its Ascend-branded AI chips using self-developed UnifiedBus interconnect protocol, a new technology also formally unveiled on Thursday.

That’s the rough equivalent of overwhelming an enemy force through sheer numbers, and its prowess is further enhanced by far faster data transmissions between individual chips — as much as 62 times quicker than Nvidia’s upcoming NVLink144 technology, Huawei claims. Meanwhile, Nvidia’s current-generation NVLink72 technology allows the company to connect 72 Blackwell graphic processing units and 36 Grace central processing units together.

Unsurprising narrative given its bloomberg. Reading between the lines, allowing up to 62x faster data transmission between chips and 107x more chips to be connected to work as one would be gamechanging.

I just hope this isn’t only useful/used for LLMs/AI and can instead also be used for other scientific workloads. The chips themselves are I’m sure optimized for low precision, high throughput tensor/matrix math, but a girl can dream of a massive cluster with boatloads of fp64 compute for doing astrophysics and biochem simulations and other things that I’m sure would bring about a greater benefit to humanity.

Also re: brute force, isn’t designing bigger and powerful monolithic chips to overcome a limitation in NVlink speed and network size just another kind of brute force?