Cambrian's market value has exceeded 280 billion yuan, and its AI chip follows the ASIC route. It launched its first NPU in 2016. It is reported that the company is developing the Siyuan 590, which directly targets Huawei's Ascend 910B and NVIDIA H100.
Although the company's current performance is difficult to match its valuation, the trend of significant growth in inventory and prepaid accounts in the third quarter shows active production investment in the future. As the demand for independent and controllable advanced process chips increases, there is a high degree of certainty that future revenue will increase.
Some institutions predict that the growth rate australia phone number list of customized AI ASIC is expected to exceed that of GPU computing.
Apart from Nvidia, Broadcom currently occupies the best ecological position in the chip field.
In addition to chips, Nvidia's trump cards for dominating the AI training market include CUDA and NV LINK. The former is a software development tool that is highly compatible with chips, and the latter is a dedicated network protocol that can provide a highspeed, lowlatency working environment.
Customized chips have freed cloud service providers from their dependence on CUDA , but Broadcom is not only excellent in ASIC, it also has a full range of networkspecific protocols and related chip IPs, which enables Broadcom to help customers develop XPU+ ultrahighspeed interconnected cluster servers.