Nvidia bets on Open AI co-founder’s new AI lab to expand computing power layout
Nvidia is investing in a low-key artificial intelligence laboratory run by former OpenAI chief scientist Ilya Sutskever, and the two parties have reached a long-term cooperation. This cooperation will not only allow the start-up to obtain more computing resources, but also help the chip giant compete with industry competitors. The two companies disclosed on Monday that Nvidia completed a large investment after gaining a rare in-depth understanding of the start-up's research and development progress; the specific financial terms of this transaction were not disclosed. The agreement between the two parties stipulates that the Safe Superintelligence (SSI) company founded by Suzkowi can use Nvidia's flagship graphics processor (GPU) hardware in large quantities, and the scale of computing power will be increased by orders of magnitude. According to people familiar with the matter, this company previously relied mainly on The tensor processor (TPU) developed by parent company Alphabet performs operations. Under the boom of the AI industry in the United States, major chip manufacturers are targeting leading AI customers through various cooperation strategies. This cooperation is a typical microcosm. In March this year, Nvidia completed another investment of the same type, also from an OpenAI veteran: former OpenAI chief technology officer Mira Mulati. Mulati founded the Thinking Machines Lab, an artificial intelligence laboratory last year. Earlier this month, the laboratory released its first AI model, relying on NVIDIA hardware to complete the training. Suzkovi is a landmark researcher who promoted the wave of artificial intelligence. Many breakthrough achievements have led the AI revolution. He co-founded Secure Super Intelligence in 2024. The start-up has a single-minded goal: to work at full speed to achieve what he calls a safe superintelligence—general artificial intelligence that the industry defines as having all the capabilities of a human without posing a threat to humanity. Top venture capital funds such as Sequoia Capital and Andreessen Horowitz quickly injected approximately US$2 billion in financing; the company's entire research and development process is highly confidential. The media previously reported that relying on Sutskvi’s top reputation in the industry, the company’s valuation reached approximately US$30 billion last year. Suzkovi said in a statement that the company's research and development focuses on research directions in the working mechanism of the human brain that have been ignored by the academic community in the past. Sutskvi was born in the former Soviet Union and grew up in Israel. He was a mathematical genius at a young age. When he went to Canada to study for graduate school, he studied under the Nobel Prize winner in economics and AI guru Jeffrey Hinton, and he became famous in the industry. In 2012, he and Hinton jointly published a blockbuster paper on computer vision, proving that relying on massive data and computing power to expand deep learning neural networks can effectively improve the intelligence level of models and directly ignite the global modern AI arms race. After that, Suzkowe joined Google and left in 2015 to co-found OpenAI. During his tenure at OpenAI, his technical ideas helped promote the birth of ChatGPT; he has long believed that as long as the amount of data and computing power for AI training continues to be expanded, the general artificial intelligence (AGI) goal set when OpenAI was founded can be achieved. Subsequently, he broke out fiercely with OpenAI CEO Sam Altman: Suzkowe participated in the vote to remove Altman; a few months after Altman was reinstated, he himself left OpenAI. After leaving his post, Sutskvi rarely spoke publicly, but he publicly questioned several times: Simply relying on existing technical routes to accumulate computing power and data may not be enough to achieve the ultimate AI goal.