This 30-year-old Chinese girl chose to challenge Nvidia's hegemony

In the world of artificial intelligence, NVIDIA's graphics processing unit (GPU) is undoubtedly one of the most popular pieces of hardware. These chips, known as GPUs, are at the heart of generative AI, and tech companies are vying to ensure that their apps like ChatGPT have enough power. This hot situation has made Nvidia's GPUs in short supply and out of stock worldwide.

Yet, in a market dominated by Nvidia founder Jensen Huang and his $2.2 trillion "GPU empire," one woman has taken a different path. She is Sharon Zhou, a 30-year-old Chinese woman, CEO and co-founder of Lamini AI.

She was only 30 years old

is challenging Nvidia's supremacy

According to reports, artificial intelligence startup Lamini AI was co-founded by Sharon Zhou and Greg Diamos in April last year, and the company focuses on providing enterprises with a customized large language model (LLM) training and creation platform using GPUs from American chip giant AMD.

▲ Sharon Chow

It is reported that Sharon Zhou is the first person at Harvard University to pursue a double degree in classics and computer science at the same time, and later received a doctorate in generative artificial intelligence from machine learning pioneer Andrew Ng at Stanford University.

She was an adjunct professor at Stanford University, where she dabbled in online teaching and angel investing. She was also invited to join the early founding team of OpenAI's rival Anthropic, which just raised an additional $2.75 billion from Amazon.

According to Sharon Chow's LinkedIn homepage, she speaks multiple languages, is a native Chinese speaker, and is fluent in French.

Choose to abandon Nvidia

Cooperate with its competitors

For the past year, Sharon Chow has been using AMD GPUs to provide hardware support for her company's product LLMs. But interestingly, AMD's chairman and CEO is Lisa Su, Huang's cousin.

Sharon Chow admits that it wasn't an easy decision to give up on the chip product that everyone is looking for in generative AI (Nvidia's GPUs). "The decision-making process is long, and it's not a trivial matter," she said. ”

Behind this decision, the company's co-founder Diamos played a key role. As a former software architect at NVIDIA, Diamos understands that GPU hardware is critical to getting the best performance out of AI models, but as he points out in his co-authored paper on the "law of scaling," software is just as important when it comes to computing power. So there's no reason why another chip company can't compete with Nvidia when they're able to build a similar software ecosystem around their GPUs.

Sharon Chow recognizes that businesses are "excited about using LLMs," but many may not want or can afford to wait for Nvidia to provide enough GPU supply to meet demand. That's another reason why she thinks AMD can make her ambitions so worthwhile. As the availability of AMD's GPUs continues to increase, Sharon Chow believes Lamini AI is the infrastructure that can provide the infrastructure to meet the rapidly soaring demand for LLMs. As a result, the company is ready to further increase its investment in AMD.

▲Sharon Chow (left) and Lisa Su

In January, Sharon Chow shared an image of the MI300X, a new AMD chip that was first introduced in December last year and hailed as "the world's highest-performing accelerator," on social media platforms.

Nvidia's Jensen Huang may now be leading one of Silicon Valley's most powerful companies, but competition is coming his way. As Sharon Chou said of AMD, "They have a real horse in this race. ”

Red Star News reporter Wang Yalin intern gold

Edited by Guo Yu and edited by Guan Li