Interface News Reporter | Peng Xin
Interface News Editor |
"You either run for food or you run to avoid being food. Often you can't tell which one is the case. Run anyway. "On May 27, 2023, NVIDIA's CEO and CEO appeared at the graduation ceremony of Taiwan University of China and delivered a speech to encourage the graduates.
Dressed in a black graduation robe with blue, white and green draped cloth inlaid with the plum blossom symbol symbolizing NTU, Huang Jenxun changed the "domineering" image of the previous black leather clothes and entered the venue under the cheers of the students. At the beginning of the speech, Huang Jenxun greeted the students in Taiwanese, "I originally wanted to speak Taiwanese with you, but the more I thought about it, the more nervous I became, I grew up in the United States, Taiwanese is not standard, is it okay to speak English today?" ”
At present, it is the highlight of NVIDIA, after delivering quarterly results that greatly exceed expectations, NVIDIA's market value has soared, approaching trillions of dollars. Under the AI frenzy triggered by ChatGPT, NVIDIA's GPUs for AI computing are in short supply, becoming the most high-profile chip company in the market.
But in this speech, Huang mentioned that NVIDIA started in the early days of micro-start-up, and the company stumbled all the way and was on the verge of bankruptcy. "I was successful before I founded NVIDIA." "But at Nvidia, I went through failures, great failures, all of which were shameful and embarrassing," he said. Many have brought us almost to our demise. ”
In his speech, Jensen Huang shared three early "failure stories" of NVIDIA, which he said played a key role in NVIDIA's three-decade journey from a three-person gaming graphics chip startup to a trillion-dollar global AI leader, ultimately shaping today's chip giants, encouraging graduates to recognize mistakes, persevere in will, and know how time changes:
The first story is that the 3D graphics chip invented by NVIDIA was first used in computer games, and after several generations of development, in order to reduce costs, it was developed by Sega, the Japanese game manufacturer at the time. However, after 1 year of development, Huang found that the industry had changed rapidly, Microsoft developed the Direct3D graphics interface on Windows 95, making the technology jointly developed by NVIDIA and Sega deviate from the mainstream of the market, Huang Jensen had to contact Sega CEO to explain this difficult situation, but he still asked Sega to pay according to the contract.
If Sega withdraws from the partnership and does not pay, NVIDIA will face the fate of bankruptcy. Fortunately, Sega agreed to pay, giving Nvidia months of respite and the opportunity to build the RIVA 128 chip, turning the company's fortunes around. "To my surprise, Sega agreed, and because of their generosity, we were able to live another 6 months." Huang said that it was Nvidia who faced up to its mistakes, stopped losses in time, and humbly sought help to survive the bankruptcy crisis.
The second story concerns CUDA (Compute Unified Device Architecture), which is now the peak of NVIDIA's market capitalization. This is both a parallel computing platform and a programming model with a range of development tools. All NVIDIA GPUs support CUDA. This means that anyone, from students to scientists, can develop software with a laptop equipped with a GPU.
In 2007, Huang decided to build CUDA into all of the company's GPUs, allowing them to process data in addition to 3D graphics. But it was an expensive long-term investment that received a lot of criticism, and it didn't pay off until a few years later when the chips began to be used for machine learning. "Our market cap is just over a billion dollars." Jensen Huang recalled, "We have experienced years of weak performance, and shareholders are skeptical of CUDA and want us to improve profitability. ”
Therefore, NVIDIA must prove the market demand for CUDA and continuously demonstrate the benefits of developing CUDA. It was not until 2014, when a scholar used CUDA for AI computing, opening the era of artificial intelligence, that CUDA was able to successfully open the market. "Fortunately, we are also discovering the potential of artificial intelligence. Years later, the AI revolution broke out, and NVIDIA became the engine. ”
Huang believes that it is precisely because of the past hardships that NVIDIA's persistence has been rewarded, "We invented CUDA for AI, and this journey has shaped our character, suffering and suffering, and it is necessary to go through pain on the way to pursue our vision." ”
The third story is about the failure of the mobile market of NVIDIA, in 2010 NVIDIA decided to enter the mobile phone market, when smartphones were booming, obviously the requirements for graphics computing will increase rapidly, but for NVIDIA, this vision did not come true, competitors poured in, learning to make mobile computing chips, while NVIDIA is learning how to develop baseband chips, although the market share is still growing, but the market competition has risen sharply.
"We made the difficult decision to give up this market. Because NVIDIA's mission is to create computers that solve 'problems that ordinary computers can't solve,' we should focus on this vision and make our unique contributions. Huang said.
Huang said the abandonment paid off, and Nvidia created a new market — robotics, which was small at the time, but now has billions of dollars in autonomous driving, robotics and a new industry.
In this regard, Huang concluded that the retreat will not be easy for the bright and successful graduates in this room. However, strategic retreat, sacrifice, and deciding what to give up are at the heart of success and are critical.
AI remained a recurring theme in Huang's speech, saying that AI will transform every job and greatly improve the performance of engineers, designers, artists, etc., and he expects that every company and everyone will learn to use AI well, and while some fear that AI may take away their jobs, experts who know AI inside will replace them.
Huang also believes that this stage is at the beginning of the era of AI technology, AI will fundamentally reshape computers, is another rebirth of the computer industry, but also a golden opportunity for Taiwanese manufacturers, he said that 1984 is a perfect graduation year, he believes that 2023 is the same, "Today is the most successful day of your life so far, because you will graduate from here." ”
"With the help of artificial intelligence, do the unimaginable and incredible." He said to the graduates present.