The rapid development of artificial intelligence has enabled in-depth communication between humans and machines. At the forum, Liu Jianwei stated that today's artificial intelligence has created new ways to utilize computing power through a new computing paradigm. The popularization of large models allows people to generate any desired form of content from simple inputs such as text, audio and images, which greatly improves the efficiency and experience of our production and daily life while further reducing costs. "It can be said that artificial intelligence today has become the intelligent infrastructure for social development."

With the increasing specialization of AI inference and training, although GPUs can reduce the energy consumption of redundant computations on CPUs, they are not built for artificial intelligence and also face the problem of high investment costs. To maximize the performance of chips, more economical, efficient and eco-friendly AI processors are needed. Based on this trend judgment, Axera launched the AxeraNeurons AI Processor.

Officially released in July this year, the AxeraNeurons AI Processor features an Operator Instruction Set and a Data Stream Microarchitecture at its core. It adopts a programmable data stream microarchitecture at the underlying layer to improve energy efficiency and computing power density. Meanwhile, its flexibility ensures the comprehensiveness of the Operator Instruction Set to support a variety of AI applications, and its mature software toolchain allows developers to get started quickly.

In terms of the co-design of hardware and software, the AxeraNeurons AI Processor chooses to "simplify". In this regard, Liu Jianwei said that in the face of the ever-growing demands of enterprises, only by disassembling and matching the core needs can the maximum efficiency of the AI processor be exerted. This also ensures the rapid iteration of the AxeraNeurons AI Processor, thus maintaining its sustainable competitiveness.

With its outstanding computing power performance and cost control, AxeraNeurons products have been widely applied in two core sectors: intelligent AIoT and intelligent driving, and cover high, medium and low computing power tiers. It helps enterprises with different computing power demands obtain AI-specific chips tailored to their business development.

Liu Jianwei introduced that the global shipment volume of AxeraNeurons AI Processors in the AIoT sector has ranked among the top. The M76H series automotive products can also perfectly support BEV multi-task models in intelligent driving applications. In the smart city sector, the AX650 series products have achieved accurate scene understanding. Enterprises can more efficiently and precisely fulfill complex demands such as text-to-image search, text-to-image generation and AI Agent, truly realizing Inclusive AI.