NVIDIA CEO Huang Renxun's latest interview! AI inference computing demand soared, Blackwell chip became the market focus

Nvidia CEO Jensen Huang said in an interview this morning, 2/27, that the global demand for AI Computing Power has reached unprecedented levels, with even Amazon AWS CEO Andy Jassy bluntly stating: 'If there were more AI Computing Power resources available now, we could sell them immediately.' This indicates that the short-term market demand for AI computing power far exceeds supply, especially with explosive growth in demand from startups.

The rise of startups is driving the expansion of AI infrastructure

Hwang In-hun pointed out that with the development of Reasoning AI and Physical AI, many AI startups are quickly rising, focusing on breakthrough technologies, and requiring a large amount of Computing Power to train and operate AI models. These startups not only accelerate the soaring demand for cloud services but also further drive the expansion of global AI infrastructure.

Data center investment surges, Blackwell chips become market focus

Looking at the mid-term, global data center investment has far exceeded last year, with Nvidia's latest Blackwell AI chip becoming a key driving force. Huang Renxun said that with these new generation data centers coming online one after another, the AI Computing Power market will continue to grow rapidly in the coming years.

The era of reasoning AI is coming, and the computational demand is 100 times that of the past

When it comes to future trends, Huang Renxun emphasizes: 'We are at the beginning of reasoning AI.' Unlike traditional AI, reasoning AI is no longer simply outputting answers, but will 'think' before responding. New models like DeepSeek, ChatGPT-4.0, Grok-3 can break down questions, search internal knowledge, and even select the best answer after self-reflection. The Computing Power required for this process far exceeds the past standards, with the computational requirements of reasoning AI being 100 times higher than traditional AI.

AI learning three stages, reasoning computing needs the highest

Huang Renxun further analyzes that the development of AI can be divided into three major stages:

Pre-training (: Similar to human learning basic knowledge, such as language, mathematics, etc.

Post-training ): AI improves its capabilities through human feedback, reinforcement learning, simulated testing, etc. The Computing Power required at this stage has far exceeded that of pre-training.

Inference (: When AI is officially running, in order to improve the quality of answers, it will first go through multiple rounds of thinking and calculation. This surge in computing power for inference is 100 times higher than traditional AI models.

The competition for AI Computing Power is intense, and the industry is entering a period of rapid growth.

The rapid surge in demand for Computing Power has significantly increased the market's reliance on efficient computing )HPC( and cloud services. With tech giants such as Nvidia and AWS stepping up their efforts in AI infrastructure, the global AI industry is entering a new phase of rapid growth.

Nvidia's financial report exceeds expectations, Bitcoin hits a low of 82K, MSTR rises more than 5%)

The latest interview with NVIDIA CEO Huang Renxun! The demand for AI inference computing skyrockets, Blackwell chips become the market focus first appeared on ChainNews ABMedia.