Saudi Arabia and the United Arab Emirates are buying thousands of high-performance Nvidia chips critical to building artificial intelligence software, joining a global AI arms race that is squeezing supplies of Silicon Valley’s hottest commodity.
Gulf powers have publicly stated their aim to become leaders in artificial intelligence, while they are pursuing ambitious plans to boost their economies. But the move has also raised concerns that autocratic leaders in the oil-rich country could misuse the technology.
According to people familiar with the matter, Saudi Arabia has purchased at least 3,000 of Nvidia’s H100 chips — $40,000 Processor Nvidia CEO Jensen Huang described it through public research institute King Abdullah University of Science and Technology (Kaust) as “the world’s first computer (chip) designed to generate artificial intelligence”.
Meanwhile, the UAE has acquired access to thousands of Nvidia chips and has developed its own open-source large-scale language model, called Falcon, at the state-owned Institute for Technological Innovation in Masdar City, Abu Dhabi.
“The UAE has made the decision it wants . . . to own and control its own computing power and talent, to have its own platform and not be dependent on China or the Americans,” said a person familiar with Abu Dhabi’s thinking.
“The important thing is that they have the capital to do it, they have the energy to do it, and they attract the best talent in the world,” the person added.
The Gulf states’ purchases of Nvidia chips through state-owned groups come as the world’s leading technology companies are scrambling to acquire scarce chips for artificial intelligence development.
Last week, the Financial Times reported that leading Chinese technology groups such as Tencent and Alibaba were seeking to acquire Nvidia’s high-performance chips.
The most advanced LL.M.s are owned by US companies, including Microsoft-backed OpenAI and Google. These groups are also major buyers of Nvidia’s H100 and A100 chips.
According to multiple sources close to Nvidia and its manufacturer TSMC, the chipmaker will ship about 550,000 of its latest H100 chips globally in 2023, mostly to U.S. technology companies. Nvidia declined to comment.
By the end of 2023, Saudi Arabia’s Kaust will receive 3,000 of these specialized chips, worth a total of about $120 million, according to two people familiar with the university’s AI lab.
By comparison, OpenAI is estimated to have trained its advanced GPT-3 model on 1,024 A100 chips (the predecessor to Nvidia’s latest chips) in just over a month.
The Saudi university also has at least 200 A100s, according to people close to Kaust, and the university is building a supercomputer, the Shaheen III, which will be operational this year. The machine will run 700 Grace Hoppers, Nvidia’s so-called superchips designed for cutting-edge artificial intelligence applications.
According to multiple sources close to the state-run university, Kaust will use the chips to build its own large language models, software that can generate human-like text, images and code, similar to OpenAI’s GPT-4, which Provides support for the popular chatbot ChatGPT.
The Saudi LL.M. was developed by Coster’s provably responsible artificial intelligence and data analysis laboratory, which is mainly composed of Chinese researchers.
According to two sources at Coster, many Chinese nationals with expertise in artificial intelligence choose to work at Coster because they cannot study and work in the United States after graduating from Chinese universities on the US Entity List.
Coster did not immediately respond to a request for comment.
Meanwhile, the UAE became the first country to set up a Ministry of Artificial Intelligence in 2017 and launched a A Guide to Generative Artificial Intelligence As part of the government’s “commitment to strengthening its global position as a pioneer in technology and artificial intelligence” and “a regulatory framework to limit the negative use of technology”.
The UAE’s Falcon model, now freely available online, was trained on 384 A100 chips for more than two months earlier this year.
“I’m really impressed with the model given the resources they use. For a while, it was one of the best models in the open source world,” says a leading AI researcher and LL.M. expert.
It also impressed high-profile venture capitalists such as Marc Andreessen, who tried to network with the team, according to an Emirati sovereign wealth investor. A spokesman for Anderson declined to comment.
According to sources from the UAE Industrial Development Authority, the UAE government has purchased a batch of new Nvidia chips to prepare for more LLM-related applications and cloud services.
Western artificial intelligence leaders and human rights experts worry that the software developed by the two countries may lack the ethical guardrails and safety features that big tech companies are trying to build into their technologies.
“Human rights defenders and journalists are frequently targeted by governments (in the UAE and Saudi Arabia),” said Yvonne McGowan, director of the European office at the Center for Democracy and Technology in Brussels.
“In fact, we know how AI can have discriminatory effects, or be used to enhance illegal surveillance. It’s a scary idea,” she added.
Saudi Arabia and the United Arab Emirates, which both manage some of the world’s largest and most active sovereign investment funds, have been buoyed by petrodollar windfalls after energy prices soared last year.
Representatives of Gulf state-affiliated funds recently approached Western AI start-ups trying to secure code and LL.M. expertise in exchange for computing resources, but they rejected their offers, the chief executives of two European AI companies said.
“We’ve got huge funding and access to data to tap our talent indirectly,” said one executive.
OpenAI CEO Sam Altman praised Abu Dhabi for its vision to recognize the importance of artificial intelligence during a visit to the region in June. During a question-and-answer session in the city’s financial district, he said the Gulf region could play a “central role in the global conversation” around emerging technologies and their regulation.
“AI was talked about before it was popular, especially in Abu Dhabi,” he said. “Everyone is jumping on the AI bandwagon now, and we’re excited about it, but we’re especially grateful to those Someone talking about AI when everyone thought it wasn’t going to happen.”
Additional reporting by George Hammond and Simon Cole
Svlook
