Before his recent headline-grabbing dismissal and quick reinstatement, OpenAI CEO Sam Altman made a short stopover in Taiwan in September to deliver a speech before traveling to the United Arab Emirates (UAE). Back then, I asked him out of mere curiosity why he was flying to the Middle East.

Altman responded, saying that the UAE has always been an “early leader and a major investor” in AI.

A recent article by Bloomberg news agency revealed that the real goal of Altman’s trip to the Middle East was raising funds for a new chip venture to compete with Nvidia Inc., which currently has a dominant position in the booming AI chip market. The report said that potential investors included sovereign wealth funds of UAE members Saudi Arabia and Abu Dhabi.

Developing AI with Government Backing

Countries in the Middle East, and Abu Dhabi in particular, have accumulated massive wealth from oil exports, which they now pour into the development of artificial intelligence. At the same time, they are buying up graphics processing units (GPU), which can process multiple computations simultaneously and are therefore suited for training artificial intelligence and deep learning models. Worldwide attention is focusing on the fact that the Gulf states are emerging as the third AI power besides the United States and China.

“They are developing AI here, using government power,” notes Hung Shih-hao, a professor at the Department of Computer Science and Information Engineering of National Taiwan University (NTU). Hung, an expert on AI supercomputers, has just arrived in Abu Dhabi to serve as a visiting professor at Mohamed bin Zayed University of Artificial Intelligence (MBZUAI).

The university is named after Sheikh Mohamed bin Zayed Al Nahyan, the ruler of Abu Dhabi and current president of the UAE who also goes by his initials MBZ. Founded only in 2019, the university is attracting top AI researchers and students from abroad with astoundingly high salaries and scholarships.

Aside from Hung, two more high-caliber AI experts from NTU are listed among the university’s faculty. Kuo Ta-wei, a professor of computer science and information engineering at NTU, teaches as a visiting professor, whereas Lin Chih-jen, a distinguished professor of Computer Science at NTU, is an adjunct professor.

Hung explains that he is most excited about the huge computing power resources at MBZUAI. Research of generative AI requires a large number of expensive GPUs, but the financial means of NTU are limited. “If you don’t have sufficient computing power, achieving innovation and breakthroughs in the key technologies for LLM is very difficult,” states Hung in exasperation.

There are two other things where the Arab Gulf states are drawing attention in the AI field.

First, in September, the Technology Innovation Institute under Abu Dhabi’s Advanced Technology Research Council (ATRC) released Falcon-180B, currently the largest open-source large language model. It is said to contain 180 billion parameters and to outperform GPT-3.5 on certain benchmarks. According to tests by Hugging Face, a website assessing model performance, Falcon-180B is the most advanced openly available model in the world, even surpassing Facebook’s Llama 2.

The other big thing highlights Abu Dhabi’s role in the quest for AI leadership from a different angle.

“Proxy of China” Conspiracy Theory

Nvidia Inc. stated in its August financial reports that the US Department of Commerce has expanded the restriction of exports of the company’s AI chips beyond China and Russia “to certain customers and regions, including some countries in the Middle East.”

Nvidia did not specify any countries. But the Daily Telegraph speculated that Saudi Arabia and the UAE, which maintain good relations with China and devote massive resources to the development of AI, were meant.

One US trade lawyer told the British newspaper that the new restrictions likely aim to prevent Chinese representative offices based in the Middle East from obtaining Nvidia chips for training AI software and then transferring the technology to China.

In other words, the U.S. government suspects that the Gulf states have probably been procuring large quantities of GPUs recently to help China circumvent the American export ban. Washington has already singled out the UAE as a “transit hub” for parallel exports to Russia.

After a closer look at the faculty list of MBZUAI, it becomes more understandable why the U.S. administration suspects that the UAE and even Abu Dhabi serve as a proxy of Chinese interests.

Shanghai-born American AI expert Eric Xing, a professor at Carnegie Mellon University’s Department of Computer Science, doubles as the founding president of MBZUAI. The ratio of ethnic Chinese among MBZUAI faculty is high. At the Machine Learning Research Department, where NTU’s Lin is teaching, 15 out of 27 listed faculty members have Chinese names.

Two of the university’s seven board members are ethnic Chinese; one is AI expert and Sinovation Ventures founder Lee Kai-fu, the other is Peng Xiao, Group CEO of Abu Dhabi state-owned AI startup Group 42 Ltd. (G42). G42 is one of the largest cloud computing companies in the Middle East. That a person with Chinese roots is made CEO of such an important company has raised suspicions in US intelligence agency circles that it might have ties to China.

Leaving aside conspiracy theories, Abu Dhabi has pragmatic reasons for developing AI. The Gulf state hopes to develop new emerging industries before all its oil reserves have been extracted to ensure a smooth transition from an oil-export-driven to an AI-driven economy.

Will we see a replay of the semiconductor debacle? Money alone does not build strategic industries, after all.

More than ten years ago, the UAE caused an uproar in the Taiwan high-tech industry. Back then, the sovereign wealth fund of the UAE bought Chartered Semiconductor from Singapore and merged California-based GlobalFoundries Inc., then the world’s third-largest semiconductor foundry behind Taiwan’s TSMC and UMC. It even had plans to build a chip factory at home to create high-tech jobs.

Abu Dhabi officials argued at the time that the capital- and energy-intensive semiconductor industry was perfectly suited for the UAE.

But subsequently, several attempts at mass production failed, the company ran up massive losses, shattering the fleeting semiconductor dream. “GlobalFoundries was a mess back then,” recalls a top executive of a semiconductor firm. “They spent huge amounts of money to recruit a bunch of mercenaries but failed to achieve anything,” says the executive in summing up the ill-fated endeavor.

In the semiconductor industry, the UAE has therefore gained the unflattering image of having “a lot of money and dumb people.” The question is now whether the UAE are headed for a similar debacle by betting on AI.

For Chu Wan-wen, an adjunct research fellow at the Research Center for Humanities and Social Sciences of Academia Sinica, whose research focuses on newly industrializing countries, the jury is still out. She says crucial is whether the UAE will be able to use the help from outside to incubate the talent and industries needed for its economic transition.

Have you read?

• TSMC in Arizona: Who wins in dispute with labor union?
• TSMC in Arizona: Navigating water shortages
• Why is Foxconn building "AI factories" with Nvidia?

Translated by Susanne Ganz
Uploaded by Ian Huang