Building Artificial Intelligence upon NVIDIA's GPU Technology

2024-02-26

"NVIDIA's AI Supercomputer is the 'AI Generation Factory' of this Industrial Revolution," said its CEO Jensen Huang during the latest earnings conference call. Huang's statement about our world being built on NVIDIA GPUs is not an exaggeration. The company controls around 80% of the market for AI data center accelerators operated by AWS, Google Cloud, and Microsoft Azure. Recently, NVIDIA's market value reached $2 trillion, adding $277 billion in stock market value on Thursday, marking the largest single-day increase in Wall Street history. The company reported revenue of $22.1 billion, a consecutive growth of 22% and a year-on-year increase of 265%, surpassing the estimated $20 billion in the latest earnings conference call. NVIDIA CFO Colette Kress stated during the earnings call, "The world has reached the tipping point of a new computing era." Huang mentioned during the recent earnings conference call, "Almost every time you interact with ChatGPT, we are doing inference. Every time you use Midtravel, we are doing inference. Whenever you see Sora or Runway generating stunning videos, NVIDIA is doing inference." He emphasized that their AI supercomputer is essentially the AI generation factory of this industrial revolution. NVIDIA has come a long way since Huang personally delivered the first DGX-1 AI supercomputer to OpenAI in 2016. Today, emerging generative AI startups like Anthropic, Inflection, and xAI heavily rely on NVIDIA GPUs, particularly RTX5000 and H100, to power their generative AI services. Earlier this year, Meta CEO Mark Zuckerberg revealed that the company is currently training Llama 3 and plans to purchase 350,000 NVIDIA H100 units by the end of this year. Additionally, the social media giant has set its sights on developing open-source AGI. It is reported that NVIDIA has invested in over 30 AI startups. Huang humbly stated, "We are honored to invest in them rather than the other way around. These are the smartest talents in the world." Kress mentioned, "Exciting companies like Adept, AI21, Character.ai, Cohere, Mistral, Perplexity, and Runway are building platforms that serve enterprises and creators." She added that emerging startups are creating LLMs to serve specific languages, cultures, and customs in many regions worldwide. NVIDIA is making significant investments in healthcare and drug discovery. NVIDIA's $50 million investment in Recursion Pharmaceuticals now provides its proprietary AI models through BioNeMo for the drug discovery ecosystem. Kress stated that in the enterprise field, NVIDIA is collaborating with "leading AI and enterprise software platforms, including Adobe, Databricks, Getty Images, SAP, and Snowflake." What's next for NVIDIA? Last year, NVIDIA faced challenges in GPU demand, but this year, according to Huang, they have improved their supply chain. "Our supply is overall improving," he said, highlighting their excellent performance in areas ranging from wafer and packaging to memory, power regulators, transceivers, networks, and cables. While shortages still exist, NVIDIA has increased the production of H200. NVIDIA plans to introduce Blackwell, a new GPU series that promises improved AI computing performance compared to the current Hopper architecture, potentially reducing the need for multiple GPUs. Huang also plans to build sovereign AI infrastructure globally. "What is happening in the United States and the West will definitely be replicated worldwide. These AI generation factories will appear in every industry, every company, and every region," said Huang. NVIDIA will host its flagship GTC conference at the San Jose Convention Center from March 3 to 21, 2024. Over 300,000 people are expected to attend the event (both in-person and virtually). "I will introduce a range of new things we have been researching in the next generation of AI," Huang said.