MiniMax has once again become a viral topic, triggered by a large language model released by Saudi Arabia.
Recently, HUMAIN, an AI firm under Saudi Arabia’s Public Investment Fund (PIF), officially launched HUMAIN M3, the first Arabic large language model in Saudi history. The chairman of HUMAIN is none other than Saudi Crown Prince Mohammed bin Salman.
However, what really drew widespread domestic attention was a detail disclosed by the authority later: the foundational base of HUMAIN M3 is the flagship model MiniMax M3, released and open-sourced by MiniMax in June 2026.
A national-level AI project funded by a sovereign wealth fund and personally led by the crown prince adopts a large model from a Chinese company as its base. This sounds somewhat like dark humor, but it also reflects from the side that Chinese large models are becoming an unavoidable technical option for many countries when building their local AI capabilities.
Putting aside the dramatic element, the Middle East’s investment in AI over the past two years has been very substantial. The most typical example is still PIF, the sovereign wealth fund with about 1 trillion US dollars under management, which planned to set up a 400 billion US dollar AI fund as early as 2024. In addition, Abu Dhabi’s MGX has just closed the world’s largest dedicated AI fund at 490 billion US dollars; Qatar Investment Authority has formed a 200 billion US dollar AI infrastructure joint venture fund with Brookfield… From Riyadh to Abu Dhabi to Doha, the entire Gulf region is betting on the same direction.
China is undoubtedly an unavoidable stop in this round of “shopping spree”. If you check the investor lists of AI companies including Zhipu AI, MiniMax, Diger Robotics, and Qianxun Intelligence, you can easily find the presence of Middle East capital. Among them, the 213 million yuan investment in Zhipu AI by Prosperity7, a fund under Saudi Aramco, is now worth 4.3 billion Hong Kong dollars (about 3.7 billion yuan) based on the current market value of 360.8 billion Hong Kong dollars, roughly 17 times the principal.
After the new energy sector, capital from the Middle East has turned its sights to the AI industry.
From Computing Power to Models: The Middle East Has Invested in the Entire AI Full Stack
How strong is the Middle East’s determination to bet on AI? A set of data can illustrate this: according to the annual report released by Global SWF, a sovereign wealth fund research institution, in January this year, global sovereign wealth funds invested a total of about 660 billion US dollars in artificial intelligence and digital infrastructure last year. Gulf funds, with Saudi Arabia, the United Arab Emirates, Kuwait and Qatar as the absolute main forces, accounted for as high as 43% of the total investment.
Entering 2026, the actions of Middle East capital have obviously reached a new level: the amount of single investment starts at several billion US dollars.
Saudi Arabia is undoubtedly the main driving force behind this magnitude jump. During the Davos Forum in January 2026, HUMAIN signed a strategic financing framework agreement of up to 12 billion US dollars with the Saudi National Infrastructure Fund (Infra) to expand artificial intelligence and digital infrastructure within Saudi Arabia; in February, HUMAIN announced its participation in the Series E financing of xAI with an investment of 3 billion US dollars, becoming an important minority shareholder of the latter.
The latest update took place during the 5th LEAP Conference earlier this month, when HUMAIN announced that it had reached more than 150 billion US dollars in project cooperation agreements, with its business covering all levels of the entire technology stack, from power supply and data centers to models and hardware.
MGX, the Abu Dhabi-based dedicated AI fund, also maintains a dense investment pace. In January, it participated in the Series E financing of Elon Musk’s xAI; in February, it led a 300 billion US dollar financing round of Anthropic; in March, it led a 1.22 trillion US dollar financing round of OpenAI; shortly after in May, it appeared in the 650 billion US dollar financing round of Anthropic at a valuation of 9.65 trillion US dollars. So far, this institution established only in 2024 has become one of the very few institutions in the world that holds shares in three AI giants: OpenAI, Anthropic and xAI at the same time.
It is worth mentioning that with its high-quality AI asset portfolio, MGX has also won the trust of investors. On July 1, MGX’s first-phase fund closed at 490 billion US dollars, exceeding the original target of 450 billion US dollars, making it the largest dedicated AI fund in history.
Except for Saudi Arabia and the United Arab Emirates, other Gulf countries have relatively small investment scales. Although they do not have the capacity for large-scale acquisitions, each has occupied a unique ecological niche.
For example, Qatar Investment Authority (QIA) established the national AI company Qai at the end of last year, and later signed a 200 billion US dollar joint venture platform with Brookfield, focusing on AI infrastructure; Kuwait Investment Authority completed its layout by acting as an LP to invest in AI funds of neighboring countries: it is not only an investor of Brookfield’s AI fund, but also one of the investors of AI Infrastructure Partners — this platform, jointly participated by MGX, BlackRock, GIP and Microsoft, aims to raise 300 billion US dollars to invest in data centers and AI computing power.
In summary, these countries have different development paths, but their combined actions have formed a complete AI value chain — from the construction of underlying computing power and data centers, to the establishment of AI companies and fund platforms, to the acquisition of equity in model companies. The Middle East has basically occupied all key positions in the industry.
The “Middle East Buyers” Behind Zhipu AI and MiniMax
China is also an unavoidable stop on the “shopping list” of Middle East capital, and taking equity in AI enterprises is the most direct way for them to enter the Chinese market.
The most typical example is Prosperity7, a fund under the oil giant Saudi Aramco.
In mid-2024, Prosperity7 led the Series C financing of Zhipu AI, with a post-investment valuation of about 3 billion US dollars. This made Zhipu AI the first large model unicorn in China with a valuation exceeding 200 billion yuan at that time, and it was also the first investment from Middle East capital obtained by a Chinese large model company.
With Zhipu AI’s successful listing on the secondary market, this investment has entered the harvest period. The prospectus shows that Prosperity7 actually contributed 213 million yuan at that time, holding about 1.2% of the shares. Based on the market value of 360.8 billion Hong Kong dollars at the close of September 17, the corresponding value has reached 4.3 billion Hong Kong dollars (about 3.7 billion yuan), with a floating profit of nearly 3.5 billion yuan.
In addition to Zhipu AI, Prosperity7’s Chinese AI portfolio also includes a series of names: vector database company Zilliz, optoelectronic hybrid computing chip enterprise Lightelligence, and AI pharmaceutical platform Insilico Medicine, covering key links in the infrastructure and model layers.
What is more noteworthy is its early layout in the application layer. As early as 2022, when embodied intelligence had not yet become popular, Prosperity7 had invested in Jaka Robotics and Fourier Intelligence; in the past two years, it has successively invested in Qianxun Intelligence, Qiongche Intelligence and Diger Robotics. Up to now, Prosperity7’s layout in China’s AI sector has covered the entire industrial chain including infrastructure, model layer and application layer.
Middle East capital also appears behind MiniMax, which was listed in the same period as Zhipu AI. During the company’s IPO, the Abu Dhabi Investment Authority (ADIA) participated in the cornerstone investment with 65 million US dollars, making it one of the institutions with the highest subscription amount among the 14 cornerstone investors.
In addition to the above cases, the UAE’s LEADX Capital led the 200 million yuan Pre-A round of Shenzhen Qiang Robotics and continued to increase its holdings in subsequent rounds; Abu Dhabi Capital Group (ADCG) signed a joint venture agreement with xKool to provide resources such as royal relationship networks, land reserves and government project access.
Apart from direct investment, Middle East capital is increasingly appearing in the list of contributors to Chinese VC funds in the identity of LPs.
In May this year, Abu Dhabi’s Arab Gulf Investment Group and Yinggang Capital announced the establishment of a global investment fund with an initial size of 500 million US dollars. It is reported that the fund mainly targets Pre-IPO, IPO and PIPE investment opportunities in the artificial intelligence and robotics sectors.
Shortly after in July, Qatar’s venture capital firm Rawdat Capital also announced that it had signed an exclusive joint investment cooperation agreement with Sinovation Ventures, thus obtaining the priority follow-up right for high-quality projects of Sinovation Ventures and jointly incubating artificial intelligence start-ups; at the same time, it also signed a cooperation agreement with 01.AI to promote the implementation of the latter’s large model solutions in the Gulf region.
This is undoubtedly a positive signal. In the past few years, Middle East capital has been the focus of attention in China’s primary market. Especially in 2023, when fundraising, investment and exit all faced multiple obstacles, “seeking opportunities in the Middle East” became a trend. However, despite the hype, there are not many GPs that have actually obtained capital from the Middle East.
Now, when the investment direction shifts from new energy to the AI market, which it is not fully familiar with but must participate in, the situation may become more open.
A Structural Transformation Related to “National Destiny”
To a certain extent, the collective large-scale investment in AI by Middle East countries is driven by lingering survival anxiety.
Back to April 25, 2016, the Saudi cabinet approved a national transformation plan named “Saudi Vision 2030”, led by Mohammed bin Salman. The situation he faced at that time was not optimistic: international oil prices had just recovered from the plunge in 2014, and Saudi Arabia’s fiscal revenue had long been highly dependent on underground oil reserves.
There is a frequently quoted self-description in this document: Saudi Arabia wants to get rid of its “oil addiction”. The path it set is also very clear: before the oil runs out, transform the country’s economic structure. There are several key directions for this transformation, and AI is placed in the most prominent position.
Shortly afterwards, the United Arab Emirates, Qatar, Oman and other countries facing the same transformation pressure also launched their own national strategies, making the development of AI the economic foundation of the post-oil era. Their goals are also very clear: Saudi Arabia hopes that AI will contribute more than 12% of GDP by 2030; the UAE’s 2031 Strategy also puts forward the goal of AI contributing 40% of GDP; Oman’s “Vision 2040” aims to increase the proportion of digital economy in GDP from 2% in 2021 to 10%…
Of course, these goals are not set arbitrarily. In addition to abundant capital, the Middle East does have its unique advantages in energy, land and location.
As we all know, AI data centers are huge power consumers, and power accounts for a very high proportion of reasoning costs. This means that whoever gets the cheapest electricity can sell Tokens at a lower price. A comparable data is: the power generation cost of Saudi Arabia’s Al Shuaibah solar project is less than 1 cent per kWh, which is about one-twentieth of that of the UK’s Hinkley Point C nuclear power project; the industrial electricity price in the Gulf region is about 0.05 to 0.07 US dollars per kWh, which is 1/5 to 1/7 of that in the UK.
At the same time, the vast desert and sparse population are very suitable for building ultra-large-scale data centers. The six Gulf countries are located at the intersection of Asia, Africa and Europe, and are important hubs of global submarine optical cables, naturally having the network advantages to connect major global markets.
Jiang Tianjiao, a researcher at the Global Artificial Intelligence Innovation Governance Center of Fudan University, pointed out in an interview with the media: “From a technical perspective, this round of AI-led technological revolution is independent of the foundation of previous industrial revolutions, which allows Gulf countries that have almost no manufacturing foundation and missed the Internet wave to ‘jump the queue’ and become a new highland of global technology, and even an important node connecting Asia, Europe and Africa.”
Of course, in addition to advantages, the shortcomings of the Middle East in developing AI are also obvious: data centers consume huge amounts of water for cooling, chip acquisition is restricted by geopolitics, and local talent reserves are insufficient — these problems cannot be solved immediately only with capital. And this precisely provides several cooperation entry points for China:
First, model base and localization. The fact that HUMAIN M3 is based on MiniMax M3 is a clear signal: the cost of self-developed large models in the Middle East is high and the cycle is long, and Chinese open-source models can be used as the base, followed by post-training with local language, culture and data in the Middle East.
Second, the computing power infrastructure supply chain. The Middle East has electricity, land and capital, but China has the most complete supply chain in the world for data center construction, servers, liquid cooling, optical modules, energy storage, photovoltaics, and power grids.
Third, applications and robotics. The Middle East has huge demand for digitalization in the finance, government, energy, medical and education sectors, and China already has cost and engineering advantages in AI applications and embodied intelligence. At the same time, Chinese companies can also leverage Middle East capital and scenarios to achieve localized implementation.
At present, the Middle East is using the money earned from oil to buy a ticket to the AI era. China is not necessarily the only seller of this ticket, but it is likely to be one of the most important co-builders.
This article is from WeChat official account “ChinaVenture”, author: Wang Manhua, published with authorization from 36Kr.