The demand for AI computing power in Gulf countries has skyrocketed, and NVIDIA and cloud giants are sensing the "gold rush" atmosphere

In terms of significant technology investments, the wealthy Gulf countries have long had a tumultuous experience. After all, the most recent investment records show that sovereign wealth funds from the UAE and Saudi Arabia invested as much as $60 billion in SoftBank's first Vision Fund, led by Masayoshi Son, but the returns have been very meager. However, in the field of artificial intelligence—where losing substantial amounts of money is a very real risk for global institutional investors—consortia from Abu Dhabi and Riyadh particularly need to develop effective investment strategy blueprints.

From the perspective of Wall Street investment banks, artificial intelligence is still in its early stages. However, there is reason to believe that the two largest economies in the world—China and the United States—will both become superpowers in artificial intelligence in the future. For the wealthy Gulf countries, they naturally do not want to miss this historic opportunity to become AI powerhouses, and they hope to transition from oil-rich nations to high-tech countries that attract global capital. However, almost all emerging AI initiatives in Gulf countries, including Saudi Arabia and the UAE, face the same infrastructure problem—namely, a severe lack of AI chips for data centers, including NVIDIA H100/H200.

Take OpenAI as an example. This startup in the field of artificial intelligence has achieved a staggering valuation of $157 billion with its AI chatbot tool ChatGPT, developed based on large language models. This valuation even surpasses that of the long-established American chip giant Intel (INTC.US). As for the undisputed leader in the AI chip sector, NVIDIA (NVDA.US), the company has launched the H100/H200 and high-performance AI GPUs based on the Blackwell architecture. These AI hardware products from NVIDIA are considered the core hardware necessary for building the AI infrastructure required for large AI models. Last year, Microsoft, as a major shareholder of OpenAI, launched the Azure OpenAI Studio cloud service based on ChatGPT's core technology, which is essentially a new AI-enhanced version of Azure, and fully integrated OpenAI's proud GPT series of large models into its flagship software products like Office, achieving growth in both performance and stock price. Microsoft has long held the title of "the world's highest market capitalization company."

According to PitchBook statistics, so far this year, AI startups in the United States have raised approximately $70 billion from global investors.

For the wealthy Gulf countries, there is a significant lack of private companies with strong technological or financing influence in the AI field. PitchBook indicates that, so far this year, venture capital firms have invested only about $700 million in AI startups in the Middle East. It is undeniable that these countries do have large national oil sovereign funds to support the development of AI startups in the region or to support tech giants like NVIDIA, Microsoft, and Google in deepening their presence in the Gulf market.

This year, the United Arab Emirates announced the establishment of a fund with a scale of up to $100 billion, MGX, which includes $30 billion in equity provided by American asset management giant BlackRock, technology giant Microsoft, and Abu Dhabi. It is understood that MGX is essentially a "technological version" of the UAE's more mature $300 billion sovereign wealth fund, Mubadala, focusing solely on rapidly investing in cutting-edge companies in the field of artificial intelligence around the world, such as OpenAI, while Mubadala invests across all sectors. Saudi Arabia is also planning to establish a similarly sized fund and has announced comprehensive cooperation with Alphabet (GOOGL.US), the parent company of American tech giant Google, to establish a super-large-scale artificial intelligence center.

Consulting firms generally believe that the UAE government is about two years ahead of Saudi Arabia in terms of investment and development in artificial intelligence, viewing MGX as an important aspect of a comprehensive AI strategy. Another key component is G42, a large company with 25,000 employees, which the Abu Dhabi government hopes to promote in establishing super-large-scale data centers and cloud computing infrastructure in the Gulf region and Africa. G42 is also developing important AI applications in healthcare, space, and data through its subsidiaries, such as M42, Space42, and Core42.

One of the most important pieces of the AI puzzle for Gulf countries like Saudi Arabia and the UAE can be said to be academia. Researchers at the Technology Innovation Institute (TII), funded by the UAE government, represent part of "Abu Dhabi Artificial Intelligence Corporation," which has developed Falcon, an AI model that has nearly the same impact as the Llama large language model developed by French AI startup Mistral and Meta Platforms, making it one of the most advanced open-source large models in the world. Consulting firms indicate that TII is key for Abu Dhabi to distinguish itself from merely being a pile of cash and hype, becoming a meaningful global AI participant.

However, almost all emerging AI initiatives in Gulf countries face the same infrastructure issues. Although Mubadala holds over 82% of the $24 billion chip manufacturer GlobalFoundries, this entity has not provided high-performance AI chips similar to NVIDIA's Blackwell, which are essential for training the most advanced large language models critical to the Gulf countries' AI ambitions Despite the extreme wealth, low taxes, and abundant sunshine of the Gulf countries reliant on oil resources, they clearly cannot create a Silicon Valley-style artificial intelligence ecosystem from scratch, composed of top experts in innovation and competition.

The good news is that Gulf countries like Saudi Arabia and the UAE are addressing these challenges by leveraging their substantial startup capital advantage, which allows them to attract global cloud computing giants to build massive data centers in the region. These cloud giants do not have to worry about potential sanctions from the U.S. government regarding cutting-edge hardware like AI chips. Companies like Amazon and Google are among NVIDIA's core customer base, prioritizing the supply of AI GPUs, thus possessing vast resources for AI training and inference computing power.

For the three major tech giants—Microsoft, Google, and Amazon—the market views them as the beneficiaries of the AI frenzy, second only to the AI chip leader NVIDIA. These three global cloud computing giants, which hold a significant market share far ahead of other participants, have been aggressively purchasing NVIDIA's high-performance AI GPUs since the AI boom began in 2023. They are focusing on developing ecosystems for B-end and C-end software applications related to AI, aiming to significantly lower the IT technical barriers for various industries to develop AI application software. Therefore, investors believe that these three tech giants, which occupy a huge share of the cloud computing market, will undoubtedly benefit from the massive software expenditure scale and cloud AI training/inference computing power expenditures brought about by global enterprises' deployment of generative AI.

Amazon AWS, Google's parent company Alphabet, and Microsoft, as large-scale cloud service providers, are undoubtedly interested in large language model (LLM) providers like OpenAI and Anthropic. However, their direct business opportunity lies in attracting large companies like G42 to connect to their immensely large data center systems, enabling a one-stop layout for building, testing, running, and driving all AI development and deployment, as well as cloud AI inference computing resources on their cloud computing ecosystem platform.

At the same time, the U.S. government hopes that emerging Gulf centers like Abu Dhabi do not get too close to China, its main competitor in artificial intelligence. This explains why Microsoft announced investments in G42 and MGX this year, and why the UAE has fully opened its data center market to U.S. large-scale cloud service providers preparing to enter the region. The implicit exchange condition is that Gulf countries like the UAE will firmly align with the U.S. AI camp rather than the Eastern camp. This exchange condition also makes it easier for large-scale data centers in the UAE and Saudi Arabia to obtain NVIDIA's AI computing resources, whether through direct purchases of NVIDIA AI GPUs or by leveraging the AI computing resources of cloud giants.

Nevertheless, even if Saudi Arabia and the UAE can secure all the necessary chips for AI development, they must decide how to utilize their AI capabilities. In this regard, Abu Dhabi seems to have a drafted plan. According to AI experts interviewed by Breakingviews in Riyadh and Abu Dhabi, the leaders of Abu Dhabi recognize that competing with Microsoft-backed OpenAI or Amazon-backed Anthropic to chase the next flagship global large language model could burn through enormous funds Gulf countries may have hundreds of billions of dollars in reserves for capital expenditures and electricity for data centers, almost equivalent to that of hyperscale data centers. However, all these participants face the risk that their large language models (LLMs) may not be attractive enough and may not be compatible with Arabic, leading to poor capital returns.

This is why the UAE is focusing on a less ambitious path. Instead of directly participating in the global AI competition, the UAE is concentrating on Arabic content and localization services, aiming to serve the Arabic market rather than challenge the dominant shares of global English and Chinese content. It is understood that over half of the web content is in English, while Arabic accounts for less than 1% of the content on the World Wide Web. The extreme scarcity of high-quality Arabic content provides Gulf countries with a reason to build their own AI databases and customized large AI models. This is why researchers from G42's Inception core department have been trekking through libraries in the region, scanning Arabic texts page by page. The scale and cost of this data processing are far lower than that of English content, ultimately used to build their own customized large model, JAIS.

The logic is simple: According to companies developing generative AI solutions similar to ChatGPT in the region, the cost of training an English large language model and then translating it into Arabic is one-third higher than the cost of training a large model using only Arabic input. Additionally, this eliminates the inherent Western biases present in most AI large models. Meanwhile, Saudi Aramco has claimed that it can provide electricity domestically at about 13% cheaper than the average U.S. data center cost of 7.8 cents per kilowatt-hour. As a result, AI participants in the Gulf region may be able to train their AI large models in a more cost-effective manner, providing a significant opportunity to sell their flagship products to the Arabic market, which has a population of 400 million, as the AI industry as a whole begins to turn a profit.

Another part of the Abu Dhabi strategy should yield more direct dividends. The UAE is more interested in improving the efficiency of its operations rather than chasing the "business-to-consumer" market of Google Search or ChatGPT. The joint venture AIQ, formed by Abu Dhabi oil giant ADNOC, G42, and Presight, has developed a dedicated AI software called "RoboWell," which enhances the labor productivity of its oil wells through autonomous adjustments based on real-time AI technology. This initiative has increased output by at least 5% and saved costs equivalent to about 30% of operating profit. Malaysia's national oil company has signed on to use a series of AIQ technologies. Meanwhile, G42 and Mubadala's health business M42 have established a medical AI large model that scored 95 on the U.S. medical exam and developed an AI solution to screen for early signs of diseases, helping to reduce radiologist costs by 20%.

According to a recent forecast by the well-known research firm IDC in its "Worldwide Artificial Intelligence and Generative AI Spending Guide," it is expected that global spending related to artificial intelligence (AI) (focusing on AI-supported applications, AI chips, and other AI infrastructure, as well as related IT and business services) will at least double by 2028 compared to current levels It is expected to reach approximately $632 billion. IDC pointed out that software or applications will be the largest category of artificial intelligence technology spending, accounting for more than half of the entire AI market in most forecasts. IDC expects the compound annual growth rate of AI software from 2024 to 2028 to reach 33.9%.

The AI competition has a sensitive geopolitical nature, and the ongoing demand for AI chips and core AI technologies from Gulf countries means that both the UAE and Saudi Arabia may find it difficult to become absolute leaders in the global AI field. However, the UAE seems to be making particularly wise decisions to become a meaningful participant. If it sticks to more prudent niche markets, AI could be an investment that is not merely built on a foundation of sand