September in Riyadh, the capital of Saudi Arabia, is swelteringly hot. As a typical desert country in the Middle East, it is hard to imagine that in ancient times, this region could support a dense population and concentrated cities solely through agricultural resources. In the last century, American oil merchants discovered vast oil resources here, propelling Saudi Arabia into modernization. Now, in the post-oil era, the country has a new vision, and in the planned industrial transformation, digitalization, especially artificial intelligence technology, has become a significant lever.
Since the launch of ChatGPT at the end of 2022, it has become increasingly difficult to overlook the presence of the Middle East in the field of artificial intelligence. Previously, the Saudi Data and Artificial Intelligence Authority (SDAIA) developed the Arabic large model ALLaM. In September, SDAIA held the GAIN SUMMIT artificial intelligence summit, attracting responsible persons from manufacturers such as Huawei, Nvidia, and Google to attend or announce cooperation on-site. Another major economy in the Middle East, the United Arab Emirates (UAE), is also frequently seen in the artificial intelligence competition. In 2023, the Abu Dhabi Technology Innovation Institute (TII) in the UAE launched the open-source large parameter model Falcon 180B, attracting global attention. That year, the UAE's artificial intelligence company G42 launched the open-source Arabic artificial intelligence language model Jais.
Advertisement
Looking at the flow of funds, the power of the Middle East is also hard to ignore. One of the six domestic large model unicorns, ZhiPu AI, received funding from Prosperity7, a fund managed by the venture capital department of Saudi Aramco, in its latest round of financing. The latest news is that the newly established AI fund MGX in the UAE hopes to participate in the latest financing of OpenAI, which could push OpenAI's valuation to 150 billion US dollars.
Temporarily shifting the focus from the competition among leading artificial intelligence companies in China, the United States, and other places, there are still some issues to be resolved in the global artificial intelligence landscape: What role does the Middle East want to play and can play in the global AI landscape? Do Saudi Arabia and the UAE have a sufficient AI domestic market demand, and what can they export abroad?
Seeking industrial transformation beyond oil
Bordering the Persian Gulf to the east and the Red Sea to the west, Saudi Arabia is the largest country on the Arabian Peninsula. When the First Financial Daily reporter arrived, the daytime temperature in Riyadh in September was over 40 degrees, and the nighttime temperature was around 35 degrees, which is almost the hottest season of the year. The temperature in Dubai, the UAE, to the east of Saudi Arabia, is similar to that in Riyadh. In terms of economic volume, in 2023, Saudi Arabia's GDP was 1.07 trillion US dollars, and the UAE's was 457.4 billion US dollars. The proven oil reserves of the two countries account for 17.3% and 5.6% of the world's reserves, respectively, ranking second and eighth in the world's oil reserves. Petrochemicals are the main industrial categories in the local area.
As the world moves into the post-oil era, both countries have been developing industries other than oil in recent years, with the digital economy, especially artificial intelligence, becoming a significant lever. In 2016, the UAE established the Dubai Future Foundation, which includes accelerators and early investment funds, focusing on fields including AI. In 2017, the UAE appointed the world's first artificial intelligence minister. In 2019, the UAE cabinet approved the "2031 National Artificial Intelligence Strategy" and the "UAE National Artificial Intelligence Plan," aiming to make the UAE a leader in the global artificial intelligence field. According to the plan, by 2031, the digital economy will account for 20% of the UAE's non-oil GDP.
Saudi Arabia is also planning industrial transformation, the most well-known of which is the "Vision 2030" proposed in 2016. In addition to outlining changes in social life, the specific goals of this vision also include economic aspects such as "non-oil government revenue increasing from 163 billion Saudi riyals to 1 trillion Saudi riyals" and "public investment fund assets increasing from 600 billion US dollars to 7 trillion US dollars." According to SDAIA, 66 out of the 96 direct or indirect goals of the "Vision 2030" are related to data or artificial intelligence. SDAIA predicts that by 2030, the contribution of artificial intelligence to the Saudi economy will reach 12%.
Whether it is planning a vision or leveraging the power of sovereign wealth funds and incubators, in Saudi Arabia and the UAE, national power is evident in AI layout. Saudi Arabia and the UAE have invested in well-known AI companies worldwide through sovereign wealth funds, and incubators established by government departments or sovereign wealth funds are incubating local startups.Data from Global SWF, a platform tracking the movements of sovereign wealth funds, shows that by the end of 2023, five out of the top ten sovereign wealth funds globally were from the Middle East, including Saudi Arabia's Public Investment Fund (PIF), the United Arab Emirates' Abu Dhabi Investment Authority (ADIA), and Investment Corporation of Dubai (ICD). Major sovereign funds in the UAE also include Mubadala Investment Company.
Among them, PIF has established a special fund called Saudi AI Company (SCAI) for investing in the AI sector. There has also been news this year that PIF is negotiating with US venture capital firm Andreessen Horowitz to establish a $40 billion fund to invest in AI. Mubadala in the UAE has invested in G42, headquartered in Abu Dhabi, which covers a $10 billion technology investment fund, an Arabic AI model, a technology talent platform, and more. Earlier this year, the UAE's Artificial Intelligence and Advanced Technology Committee also established the technology investment company MGX, with Mubadala and G42 as founding partners, investing in areas including AI infrastructure, semiconductors, AI core technology, and applications.
This year, both Mubadala and MGX have appeared in the financing news of the world's top large model companies. Earlier this year, a subsidiary of Mubadala invested nearly $500 million to acquire a stake in Anthropic, one of the strongest competitors of OpenAI. If MGX successfully participates in OpenAI's new round of financing this time, Emirati capital will emerge behind the two most watched large model companies in the world.
According to data from Pitchbook, an institution in the United States that specifically tracks the venture capital industry, the scale of investment funds from Middle Eastern sovereign wealth funds in AI companies has increased fivefold in the past year.
For startups, the UAE and Saudi Arabia support through incubators. In 2017, Hub71, an incubator for the early stages of investment projects, was established in Abu Dhabi, UAE, with the support of Mubadala. Last year, SDAIA, the Saudi National Technology Development Program, and AI platform New Native launched the generative AI accelerator GAIA. Data shows that GAIA has incubated 50 AI startups in its first three batches.
After planning AI layout, the AI ecosystem of the UAE and Saudi Arabia has changed significantly. Zhu Xiujian, founder and CEO of Dhow Holding, a comprehensive investment company in the Middle East, told reporters that Middle Eastern funds have been investing in the technology sector for a long time, such as participating in top Silicon Valley projects through dollar funds. However, when he first came to the Middle East in 2015, he found that almost no one was doing AI locally. When the Middle East was considered a place that could develop an AI career, it was around 2018. Among them, the UAE began to truly invest in local AI projects in 2017, and in 2018, Zhu Xiujian's CV (computer vision) company, Machine Colony, was acquired by a family office in Abu Dhabi. Saudi Arabia's large-scale layout of AI was around 2020, launching accelerators and incubators.
The Middle East's "money power" has already provided considerable support to enterprises. Zhu Xiujian told reporters that when Saudi Arabia and the UAE support local AI companies, the requirements for shareholder backgrounds are not high, only requiring institutions to be located locally and to operate businesses locally. His CV company and three other startups were selected by the Dubai Future Foundation as the first batch of incubation projects in 2016. The main accelerators in Saudi Arabia include GAIA, and early projects can receive 250,000 riyals (about $67,000) as long as they are selected.
Demand is hidden in urban construction.
It takes about 40 minutes to drive from the Riyadh diplomatic quarter to King Abdullah Financial District (KAFD). The closer you get to KAFD, the taller and more avant-garde the buildings in the field of view become. KAFD has been under construction for more than ten years, and after the project was acquired by PIF in 2018, it has developed into a business district with more than 90 large building complexes. There are also some roads being repaired near KAFD, and taxi drivers need to detour around the roads under construction before finding a route, which takes much longer than originally expected. Not only KAFD, but the entire urban background of Riyadh also includes building and road construction. The route and duration estimated by the taxi software may not be accurate, and it is not uncommon to encounter traffic jams during peak hours and detours.
The First Financial reporter met Huang Jigong, President of SenseTime's international business, in SenseTime's office in KAFD. Since arriving in Saudi Arabia in 2018, SenseTime has witnessed and participated in the local smart city construction and has carried out AI business in it. Huang Jigong told reporters that traffic in Riyadh is still "troublesome" until now, and SenseTime is still discussing with local parties on how to solve it. "There are many newly built buildings, and it is necessary to predict how the increased passenger flow will affect traffic after the increase." When asked about local AI demand, Huang Jigong told reporters. He said that Saudi Arabia is also developing cultural tourism at present, and there is room for AI to play, such as digitalizing hotel check-ins.In 2018, when SenseTime was considering entering the Middle East market, it judged that both Saudi Arabia and the United Arab Emirates were relatively open markets. Huang Jigong and several of the company's founders visited Saudi Arabia for five or six trips to conduct surveys. In the initial plan, the scenarios where computer vision (CV) could be applied in Saudi Arabia included building operations and maintenance. For instance, due to the abundance of desert sand, building windows need to be cleaned frequently. However, there is a shortage of talent related to building operations and maintenance in Saudi Arabia, and AI companies can collect photos to determine how buildings should be maintained. In 2021, SenseTime established a joint venture, SenseTime MEA, with the Public Investment Fund (PIF), with SenseTime and PIF holding 51% and 49% of the shares, respectively. Subsequently, the Saudi Computer Vision Artificial Intelligence Institute (SCAI) invested 776 million Saudi Riyals (approximately 1.429 billion RMB) in this joint venture and used it as the main entity to build a high-end artificial intelligence laboratory locally. SenseTime's current cooperation with PIF covers areas such as smart cities, intelligent transportation, and smart cultural tourism.
In the Middle East, SenseTime has been assisting governments or enterprises with digital transformation. Previously, it mainly provided AI platforms and algorithms to customers, but now its local business is also transforming, shifting more towards end-to-end project acceptance. Over the past few years, the Middle East has been one of the fastest-growing regions in SenseTime's international business segment. Last year, SenseTime had 40 employees in Saudi Arabia, but now there are more than 80, with an expected number exceeding 100 by the end of the year. The demand for R&D and sales talent is increasing. The local market has become more vibrant, and SenseTime's employees have been busier this year.
"Five years ago, when talking about AI here, everyone was not very clear about how to use it. The most familiar application was CV in security and facial recognition. Now, there is a growing understanding of how AI can help improve operational efficiency or help make money in various scenarios. This is a very obvious change. Now, projects are more pragmatic and have timetables," Huang Jigong told reporters, adding that sometimes the timetables for PIF's national-level projects are very urgent, and there is a strong sense of crisis, which is good for business development.
With the penetration of internet products, AI is also entering the social life of the Middle East.
Zhao Jianan, the person in charge of Tencent Cloud's overseas entertainment, told reporters that many Chinese companies went to the Middle East to do social live streaming business in 2015. In the past, some offline entertainment activities in the local area were subject to certain restrictions, and people preferred online socializing. After several years of rapid growth in online socializing, the local market is now relatively mature. "In the first half of this year, I went to the Middle East and found that the market is not quite the same as originally understood. I originally thought that users in the Middle East were more closed in social interactions, but in fact, they really like socializing. Some people may have accounts on multiple platforms and are more active on online media," said Zheng Dong, the technical person in charge of Tencent Cloud's overseas business. Data from social media agency We Are Social and Hootsuite in 2020 showed that the UAE has an average of 10.5 social media accounts per person, ranking first in the world.
With the overseas expansion of domestic audio and video companies, Tencent Cloud is also conducting business in the Middle East. Zheng Dong told reporters that the AI functions of Middle Eastern audio and video products include gift generation and some cool 3D visual performances, involving more AI inference and more AI applications. AI applications also include using 3D digital humans for tour guides and e-commerce shopping. Since last year, Middle Eastern customers' demand for AI inference has increased.
In March of this year, Tang Daosheng, Senior Executive Vice President of Tencent Group, said that Tencent's investment in the Middle East region is still in its infancy, but there is a lot of potential, and investment will be increased. Tencent is investing in the AI field, and some of its customers in the Middle East already have AI needs.
Among Chinese-funded enterprises, Huawei and Alibaba have also come to the Middle East. In 2023, Huawei Cloud's Riyadh node went online. In September, at the GAIN SUMMIT World Artificial Intelligence Summit hosted by SDAIA, Qi Xiao, President of Huawei Cloud Middle East and Central Asia, said that Huawei is providing end-to-end artificial intelligence solutions. Huawei Cloud previously revealed that the Saudi government has used Huawei Cloud to develop an Arabic large language model, supporting multiple AI applications.
From existing cases, the demand for AI in the Middle East is generated with the digital needs of governments and enterprises and urban construction. From the perspective of AI To C facing local consumers, the populations of Saudi Arabia and the UAE are 32.18 million and 10.24 million, respectively, and the global population using Arabic is about more than 400 million people. The purpose of developing AI is to improve the operational efficiency of governments and enterprises, improve urban life, or get closer to consumers. Whether it is mainly aimed at domestic demand or expanding its influence to the Arab world and even globally, these two Gulf countries that regard AI as an important strategy need to answer these questions.
If AI is to get closer to consumers, the短板 of network infrastructure may still need to be made up. A cloud practitioner told reporters that there is a weak network environment between regions in Saudi Arabia, connected by deserts. Saudi Arabia is also increasing the construction of network infrastructure. According to the plan, by 2030, Saudi Arabia will achieve 90% high-speed broadband coverage in densely populated areas.How Big Can AI Ambitions Be Supported?
Having lived and worked locally for 9 years, Zhu Xiujie has a deep impression of the gradual establishment of AI territories in Saudi Arabia and the United Arab Emirates (UAE). He started working in Dubai in 2015, engaging in policy research in the technology policy team of the UAE Prime Minister's Office and participating in the establishment of the first government-guided technology accelerator and the Dubai Future Foundation. His current investment company operates in both the UAE and Saudi Arabia, with investment fields including AI, e-commerce, etc.
In addition to having a personal experience of the "money power" construction, Zhu Xiujie has also noticed the shortcomings of the AI ecosystem in Saudi Arabia and the UAE. "In the UAE and Saudi Arabia, 'small money' and 'big money' are not difficult to obtain, including early seed rounds, angel rounds of projects that want to get financing of 100,000, 500,000 US dollars, and enterprises that want to get C round, D round financing during the rapid growth period, government-backed incubators and sovereign funds can enter respectively. However, it is difficult to obtain money in the middle stages of A round and B round, which is just the stage when enterprises need to spend a lot of money to seek growth." Zhu Xiujie told reporters that when facing AI, even countries with huge cash like Saudi Arabia and the UAE may not be able to depict a clear future and make precise investments.
"AI is like electricity, and it is not clear how to 'use electricity'. This wave of generative AI is not driven by solid business scenarios. Whether in Saudi Arabia and the UAE, it can be almost certain that AI projects will not make money. No matter how strong the talent background of an AI company is, it may still be a negative asset. In the past, Middle Eastern capital invested in industries such as new energy factories, and the finances would be relatively clear, but AI is not the case. AI still lacks sufficient domestic demand, and the investment of these two countries in AI is more like a bet on the richness of future industries. This may be the reason why their considerations are more complex and cautious." Zhu Xiujie said. According to the information he has learned, there are also differences in the depiction of the AI panorama between the UAE and Saudi Arabia. Enterprises in the UAE are more international, AI technology can cover overseas areas, and the positioning of AI is more pragmatic. In the future, it may point to being a regional leader. Saudi Arabia regards AI as infrastructure like water and electricity, and its ambitions at the current stage may be greater.
In terms of the upstream infrastructure of AI, Saudi Arabia intends to layout semiconductors domestically. The company Alat under the Public Investment Fund (PIF) plans to invest $100 billion by 2030, with targets including the establishment of semiconductor factories, etc. Alat plans to produce 1 million wafers by 2030. In June this year, Saudi Arabia established a National Semiconductor Center, with the goal of attracting more than 10 billion Saudi riyals (about 260 million US dollars) in investment to promote domestic electronic chip design and the establishment of 50 professional companies. The UAE also has certain considerations for the development of semiconductors, but it may mainly invest rather than build factories locally. Previously, when AMD split its manufacturing department to become the later semiconductor foundry GlobalFoundries, the Abu Dhabi Technology Investment Company (ATIC) was a shareholder of GlobalFoundries.
"Introducing semiconductors from the very upstream infrastructure, but whether the top semiconductor companies are willing to fully enter the local area is a question. If the layout of AI is too large, it may not be able to support the funds. From this perspective, Saudi Arabia's AI strategy still has the possibility of adjustment, and the funds may be inclined to the new future city project NEOM." Zhu Xiujie said that there was news before that due to some project delays and excessive expenses, the Saudi government has adjusted the budget for the main core projects of Vision 2030 this year. Saudi Arabia may also extend the project cycle while maintaining the total budget unchanged, and reduce the expenditure on large projects such as NEOM before 2030.
Looking at the different situations of Saudi Arabia and the UAE, Huang Jigong believes that after the release of the "Vision 2030", Saudi society has become more vibrant. In terms of economic development, Saudi Arabia is similar to the domestic situation more than 20 years ago, and young people can see that future living conditions and career prospects are improving. In contrast, the UAE's economy has developed for two to three decades, forming a more mature business ecosystem. Some multinational companies first entered the UAE, and then came to Saudi Arabia to set up regional headquarters as the Saudi market developed.
After planning and investing in the AI industry, these two economies in the Middle East need to make steady progress and exert greater influence in the field of AI. The shortage of talent and infrastructure is a challenge that needs to be addressed.
Zhu Xiujie believes that the "sticking points" for the development of AI in the Middle East include related talents, such as engineers, etc. Huang Jigong told reporters that the年轻化 of the Saudi population is obvious, with about 70% of the population under 30 years old. There is no lack of management talents and they are very international. Many government or enterprise executives have studied in Europe and the United States, and the executives of Saudi Aramco are also very professional. The challenge faced by Saudi Arabia is that the economic focus has been on energy in the past, with more talents in the energy industry. Now, it needs to transform to the technology industry, and there is a lack of AI talents and engineering talents.
"However, it is now obvious in Saudi Arabia that the way of thinking of young people is changing, with a greater sense of urgency. In another 5 or 10 years, talents will be sufficient." Huang Jigong said.Saudi Arabia and the United Arab Emirates (UAE) have already initiated efforts to cultivate local artificial intelligence (AI) talent. Within the year, the UAE's National AI Program launched its fifth AI project, with goals that include nurturing creative talent for new fields in the future. In Saudi Arabia, SenseTime had previously implemented an AI education program with the Saudi Data & AI Authority (SDAIA), providing AI courses for local teachers and students. During this year's GAIN SUMMIT, SDAIA also signed a Memorandum of Understanding (MoU) with Huawei, initiating a project for the cultivation of information and communication technology (ICT) and AI talent.
In terms of AI infrastructure, Saudi Arabia and the UAE still have much work to do.
Data from market research firm DC Byte shows that, as of the end of 2023, the UAE's data center capacity (Live Capacity) was 235 megawatts, and Saudi Arabia's was 123 megawatts. With data centers under construction and planned, both countries are expected to exceed a data center capacity of 500 megawatts. Since 2020, the demand for data centers in Saudi Arabia and the UAE has been rapidly increasing. However, compared to other regions in EMEA (Europe, the Middle East, and Africa), the data center capacity of Saudi Arabia and the UAE is not large. London and Frankfurt alone have capacities exceeding 1000 megawatts and over 500 megawatts, respectively. With data centers under construction, committed to construction, and in the early stages, the estimated data center capacities for these two cities are expected to exceed 3.5 gigawatts and 2.5 gigawatts, respectively.
Large models, which have significant computational requirements, have hard demands for computational infrastructure. Currently, SenseTime uses more small models in the Middle East. In addition to having overseas supercomputing centers in Hong Kong and Singapore, SenseTime plans to build a computing center in Saudi Arabia to promote the application of large models. Reporters have learned that other domestic cloud vendors that have launched large models also plan to build data centers in the Middle East. Some international leading cloud vendors also plan to build data centers in the Middle East. This year, Amazon Web Services (AWS) plans to launch an AWS infrastructure region in Saudi Arabia by 2026, as part of its long-term commitment, with AWS planning to invest over $5.3 billion in Saudi Arabia.