Everything About AI
NextFin News - OpenAI chief executive Sam Altman is traveling to the United Arab Emirates this week to sign energy and investment agreements, a trip that places the Gulf state's sovereign capital and power grid at the center of the world's most expensive artificial-intelligence infrastructure race. The visit, confirmed by the Abu Dhabi Media Office, pairs the two resources that have become the binding constraints on AI expansion: cheap, abundant electricity and patient sovereign money. Altman met Crown Prince Sheikh Khaled bin Mohamed Al Nahyan for talks on energy and investment cooperation, according to a statement from the Abu Dhabi Media Office on Monday.
The Deal Behind the Headline
Altman's Abu Dhabi stop is the latest leg of a global fundraising and supply-chain campaign that has taken him across Asia and the Middle East in search of financing and manufacturing partners capable of meeting OpenAI's insatiable demand for computing capacity. The timing is deliberate: OpenAI is pursuing a funding round that people familiar with the matter have placed at roughly $40 billion to finance model development and the Stargate data-center plan, while separately exploring a larger capital raise that could reach $100 billion and value the company at as much as $830 billion if fully subscribed.
The UAE leg follows a whirlwind Asia tour in which a partnership with Japan's SoftBank for AI services was announced. SoftBank is in talks to lead a funding round of up to $40 billion in OpenAI at a valuation of $300 billion, including the new funds, which would rank among the largest single fundraises ever for a private company. The juxtaposition is telling: a $300 billion valuation in Tokyo, a potential $830 billion ceiling in a round that may not close until the end of the first quarter at the earliest, and terms that people close to the discussions caution could still change.
The capital ask is only half of the equation. The other half is power. OpenAI and its partners have committed to a Stargate cluster in Abu Dhabi with a planned capacity of 1 gigawatt - among the most powerful data-center complexes in the world - with 200 megawatts expected to come online in 2026. The campus sits inside a broader 5-gigawatt UAE-US AI zone covering roughly 10 square miles, the largest AI infrastructure project of its kind outside the United States. Stargate partners have pledged to spend $100 billion on the servers that provide AI's computing muscle, ramping to $500 billion over four years.
The combination is the point. An AI data center is, in operational terms, a power plant with networking gear attached. A 1-gigawatt cluster draws enough electricity to supply roughly 750,000 average U.S. homes, and it must run at 99.999% uptime. Securing a sovereign partner that can deliver both financing and guaranteed power converts a capital problem into an execution problem - which is precisely what Altman is in Abu Dhabi to negotiate.
The Real Constraint Is Watts, Not Dollars
For two years the AI build-out was framed as a race for chips. Nvidia's GPUs were the scarce input, and nearly every major announcement led with chip counts. That framing is now outdated. The binding constraint has shifted upstream to electricity generation and grid interconnection.
The mechanism is straightforward. Training frontier models and running inference at scale requires clusters that operate continuously at high utilization. Each successive generation of models has demanded more compute than the last, and compute demand translates almost one-for-one into power demand. A data center can be financed in weeks; a gas-fired power plant or a grid interconnection takes years. When the scarce input moves from chips to kilowatt-hours, the competitive advantage shifts from the companies that can buy silicon to the jurisdictions that can guarantee power.
The Gulf holds structural advantages on both sides of that equation. The UAE produces some of the lowest-cost electricity in the world, anchored by cheap natural gas and a growing nuclear and solar base. Abu Dhabi aims to generate 60% of the emirate's electricity from renewables and nuclear by 2035. Projections put regional data-center power demand at 15 to 20 terawatt-hours by 2026, and the UAE's G42 Stargate campus alone is set to bring 200 megawatts of load online this year before scaling to the full 1-gigawatt cluster.
That is why the visit is structured around energy deals as much as investment deals. OpenAI is not shopping for a checkbook; it is shopping for a location where capital and power arrive together.
"There's a reason that everyone from Sam Altman to Elon Musk comes to this region and talks to us. The US, as a home base, cannot provide everything they need."
Peng Xiao, chief executive of G42, the Abu Dhabi technology conglomerate leading the campus construction, made that observation in an interview earlier this year. It captures the bargain: the Gulf offers speed and scale that the U.S. grid, constrained by permitting and interconnection queues, currently cannot match. G42 has drawn funding from Microsoft and signed deals with Nvidia, positioning itself as the Gulf's bridge to American AI supply chains.
The scale of the regional build-out extends well beyond OpenAI. The Middle East data-center construction market is projected to grow at a compound annual rate of roughly 26% from 2025 to 2031, driven by AI-ready facilities and liquid-cooling adoption expected to rise from 41% of deployments in 2025 to 54% by 2031. Across the wider Gulf Cooperation Council, more than 174 active and planned data-center projects are being tracked, with a combined value exceeding $93 billion. In that context, Altman's visit is less a singular negotiation than a checkpoint in a regional reallocation of capital.
Sovereign Capital Is Patient in a Way Venture Capital Is Not
The second transmission channel is the nature of the money. OpenAI's infrastructure plan is measured in the hundreds of billions, with internal planning scenarios reaching into the trillions. No traditional venture fund can write that check on a five-to-ten-year payback. Sovereign wealth vehicles can.
Abu Dhabi's AI push is led by state-backed G42 and MGX, an investment group established in March with Mubadala Investment Co. and G42 as founding partners and a stated goal of eventually exceeding $100 billion in assets. Mubadala itself manages roughly $330 billion and reported deploying 119 billion dirhams, or about $32 billion, in a single year while growing its assets under management to 1.2 trillion dirhams. MGX is overseen by Sheikh Tahnoon bin Zayed Al Nahyan, the UAE's national security advisor and brother to the country's president - a structure that places AI investment inside the national-security apparatus rather than under a purely financial mandate.
That distinction matters for deal terms. A sovereign investor optimizing for strategic positioning - jobs, technology transfer, regional AI leadership - can accept returns and timelines that a fund optimizing for internal rate of return cannot. For OpenAI, which needs capital that will not flinch when a $500 billion, four-year build-out hits inevitable delays, that is the right kind of money.
The UAE has already shown it can move quickly. MGX participated in a $6.6 billion secondary share sale of OpenAI stock in October that valued the company at $500 billion. The same vehicle is now being positioned as a lead partner for the next stage. The progression - $500 billion in October, a reported $300 billion figure attached to the SoftBank-led round, and an $830 billion ceiling if a larger raise is fully subscribed - reflects a market still searching for a stable price on the world's most valuable private company.
The Geopolitical Hedge Cuts Both Ways
Altman's strategy is also a hedge against concentration risk. The Stargate UAE project is the first major deployment of the initiative outside the United States, part of what OpenAI calls its "OpenAI for Countries" effort to help allied governments build sovereign AI capability.
"This is the first major milestone in our OpenAI for Countries initiative - our effort to work with allies and partners to build AI infrastructure around the world," Altman said.
Under the partnership, the UAE is set to become the first country in the world to enable ChatGPT nationwide, giving people across the country access to OpenAI's technology. OpenAI has stated that the project has the potential to provide AI infrastructure and compute capacity within a 2,000-mile radius, extending the geographic reach of its compute beyond North America.
But the hedge introduces its own exposure. U.S. officials have long voiced concern over the UAE's warm relationship with China and the risk that American technology could be shared with Beijing. Export controls on advanced chips are the enforcement mechanism, and any perception that Gulf-hosted infrastructure weakens that perimeter would invite tighter restrictions - which would defeat the purpose of locating capacity abroad.
There is also physical risk. Critical infrastructure in the region has become a potential target amid regional conflict, forcing U.S. technology companies to price war risk into assets they had assumed were insulated from it. A data center is not easily relocated once built, and the concentration of American AI infrastructure in a region where pipelines, airfields, and power plants have long been military targets represents a new category of operational exposure.
The Counter-Thesis: Leverage Theater, Not a Binding Deal
The strongest case against reading too much into the trip is simple: nothing is signed yet, and the numbers keep moving. OpenAI's fundraising is in the early stages, people familiar with the matter have said, and it remains unclear whether investor demand will be sufficient to reach the targets. Terms can change. The $40 billion figure, the $100 billion figure, and the multi-trillion-dollar internal scenarios are all moving targets sourced from anonymous informants, not commitments.
There is also a serious competitive counterweight. China's DeepSeek demonstrated earlier this year that frontier-level capability can be reached at a fraction of the cost, which undermines the premise that only the highest-spending lab wins. If efficiency gains outpace spending, the company that raises the most money does not necessarily win the race - it just loses more slowly.
Finally, the Gulf is not the only option. The United States still offers the deepest capital markets, the most advanced chip supply chain, and the strongest rule of law for intellectual property. Moving core infrastructure abroad trades one set of risks for another.
The counter-thesis has force, but it mistakes the nature of the transaction. Altman is not choosing between the Gulf and the United States; he is diversifying across both. The U.S. remains the home base, and the Gulf becomes swing capacity - the place where projects can be permitted and powered quickly while domestic interconnection queues remain a bottleneck. The risk is not that the UAE replaces America; it is that the UAE gains enough leverage over OpenAI's roadmap to extract concessions on pricing, localization, and governance.
What Would Prove the Bullish Read Wrong
The specific falsifying signal is straightforward: if the 200-megawatt first phase of Stargate UAE does not come online in 2026 as planned, or if the UAE funding does not close within the next two quarters, the "capital and power arrive together" thesis breaks. A missed power milestone would indicate that even the Gulf's advantages cannot outrun the physics and permitting of AI infrastructure - and that the bottleneck is deeper than geography.
Who Wins, Who Is Exposed, and What Comes Next
The near-term beneficiaries are clear. Oracle, Nvidia, and Cisco are named partners in the UAE cluster; Oracle has separately been reported to be planning roughly $40 billion in Nvidia GB200 chip purchases for OpenAI data centers. Every gigawatt of AI capacity committed translates into orders for GPUs, networking gear, and cloud orchestration. The exposed parties are the labs that cannot match this capital intensity: if the cost of staying competitive rises to hundreds of billions, the field narrows.
Short term, the story is sentiment and deal momentum - another announcement, another partner, another valuation mark-up. Medium term, it is about which facilities actually deliver power and compute on schedule; that is where the UAE's execution record will be tested against U.S. interconnection delays. Long term, this is structural: AI infrastructure is becoming a sovereign asset class, funded by states rather than markets, and located where power is cheapest and permitting is fastest.
Base case: the UAE round closes, the 200-megawatt phase powers up in 2026, and OpenAI uses the Gulf as swing capacity alongside U.S. expansion. Upside case: the full 1-gigawatt cluster comes online ahead of schedule and the funding round reaches the higher end of the reported range, cementing the UAE as OpenAI's primary non-U.S. hub. Downside case: export-control friction or regional security concerns delay the project, and capital stays concentrated in the United States.
Altman's trip is not really about the UAE. It is a referendum on whether the United States can power the AI boom it started - and so far, the answer from the companies building it is that they cannot wait.
Data as of November 2, 2026. Fundraising figures are based on people familiar with the matter and remain subject to change.
Explore more exclusive insights at nextfin.ai.
