Saudi Arabia and the United Arab Emirates (UAE) are pursuing aggressive AI strategies by securing thousands of high-performance Nvidia chips.
Reportedly, Saudi Arabia, through the public research institution King Abdullah University of Science and Technology (KAUST), has purchased a minimum of 3,000 of Nvidia’s H100 chips, each priced at approximately $40,000.
The UAE isn’t far behind, also acquiring thousands of Nvidia chips, and the Technology Innovation Institute of Abu Dhabi has already created a 40-billion parameter open-source large language model (LLM) named Falcon. Falcon is a high-performance LLM that once sat at the top of Hugging Face’s leaderboard.
An analyst told the Financial Times, “The UAE wants to own and control its computational power and talent, develop their platforms, and remain independent from the influence of major players like the Chinese or the Americans. With the capital and energy resources at their disposal, they’re also attracting the top global talent.”
The AI arms race is heating up globally, with leading tech giants vying for control of as many high-end chips as they can get their hands on.
Nvidia is already poised to ship around 550,000 H100 chips worldwide in 2023, primarily to US-based tech firms. Chinese firms also placed a significant order to the value of around $5bn as the US government further moves to restrict tech exports to the country.
Meanwhile, Gulf universities and research institutions are launching plans to launch a supercomputer named Shaheen III, which KAUST will use to train an LLM similar to OpenAI’s GPT-4.
Some have criticized the potential misuse of AI by authoritarian states. For instance, Iverna McGowan, director of the Center for Democracy and Technology’s Europe office, pointed out that AI could enhance government crackdowns on human rights defenders and journalists.
The AI race is not two-sided
The Gulf states are challenging the traditional dual superpower dynamic established by the US and China, though China’s purchase of some 100,000 Nvidia chips – albeit weaker A800 variants – is an extreme example of how far businesses will go to secure AI hardware.
However, businesses in Saudi Arabia and the UAE often enjoy strong government backing, which will be an asset for developing a cohesive AI ecosystem.
The future leaders of AI may be dictated in part by innovation and partly by who can secure the most AI chips – as a shortage might be imminent.
After all, Nvidia is more-or-less out on its own here – for how long can they sustain it?