
Last month’s AI deal between Donald Trump and the Gulf states signaled a strategic shift. The United States, long focused on locking down access to advanced chips and AI infrastructure, is now beginning to share that power with less selectivity. With a landmark agreement to allow Saudi Arabia and the UAE to import over one million U.S.-made AI chips for use in new data centers, Trump has opened the door to a new phase of tech geopolitics.
This move upends the Biden administration’s “AI diffusion” strategy, which created a tiered system of export controls for AI tech, limiting exports to adversaries and restricting others to partial participation. By reclassifying Gulf allies in practice as trusted partners, Trump is betting that alignment is more important than control. The question now is whether this will strengthen U.S. leadership in the AI age or accelerate the erosion of America’s technological dominance.
Four Pillars of AI Power
To understand the AI strategy, you have to understand its components. All AI models, even the most powerful, rest on four pillars: chips, software, data, and infrastructure.
Chips
Chips are the physical brains of AI systems. Without them, no model runs, no data gets processed, and no prediction gets made. The U.S. dominates chip design through firms like NVIDIA, AMD, and Apple. And the world’s most advanced chips are made by TSMC, a Taiwan-based manufacturer.
Washington has gone to great lengths to keep these chips out of Chinese and Russian hands through sanctions, export controls, and diplomacy. This deal represents a break in that logic. While Gulf states aren’t adversaries, they’ve never been treated as full participants either. That may be changing.
Software Models
AI doesn’t just need hardware. It also runs on software known as large language models (LLMs). These are statistical engines that “learn” associations between words and phrases so well they can “predict” the most accurate response to a prompt. But these models have become so complex that even their developers don’t fully understand how an AI arrives at a particular output.
When you hear experts call these models “black boxes,” that’s not metaphorical. It’s literal. This makes regulation tricky: how do you govern outputs you can’t predict? Trump’s move brings this challenge to the fore. The chips being exported are only part of the picture. The software and intellectual property that surrounds them, much of it still U.S.-based, could follow.
Training Data
Think of data as the fuel that trains AI systems. Without huge datasets of digital information, AI can’t “learn” anything. But where this data comes from, and who has the right to use it, is now hotly contested. Legal fights over copyright, privacy, and scraping are underway in courts around the world.
What is owed publishers when their words are used to build a system that generates profits for another company? Allowing Gulf states to build training centers with U.S. chips raises new questions: Should American data sets, even anonymized ones, be shared abroad? Will models trained in the UAE end up incorporating sensitive U.S. inputs? Who gets to decide what is safe?
Infrastructure
None of this works without power. AI needs large datacenters filled with servers to function. These facilities consume land, capital, and huge amounts of electricity. And this is where the Gulf excels. Countries like Saudi Arabia and the UAE have cheap energy, sovereign wealth funds, and the political will to build fast. They don’t need to invent the tech; they just want to host it. The Trump deal allows exactly that. U.S.-aligned chips, running in Gulf-funded infrastructure, training models that may or may not follow American norms. Whether this becomes a strength or a vulnerability depends on what happens next.
Supply Chains, and Strategic Leverage
This isn’t the first time the U.S. has had to grapple with the collision of physical and digital power. For example, China dominates the dirty, low-margin parts of refining critical materials to manufacture chips, while the U.S. and allies specialize in ultra-pure inputs and advanced manufacturing. The AI story follows a similar pattern: geopolitical value flows not just from who holds the materials, but from who controls the stages of production and deployment.
The Gulf states are trying to skip the queue. They lack the intellectual property, the research universities, and the tech breakthroughs. But they have capital. They have land. And they have a foreign policy strategy that increasingly revolves around becoming indispensable to great powers. whether the U.S., China, or both, to head off.
The AI Diffusion Debate
Biden’s AI diffusion policy aimed to create a controlled system. It divided the world into three tiers:
Tier 1: Trusted allies like the UK, Japan, and Germany could access U.S. AI chips freely.
Tier 3: Adversaries like China and Russia were outright banned.
Tier 2: Everyone else—Gulf countries included—could access some technology, under tight restrictions.
The logic was simple: keep the most powerful tools close and avoid enabling competitors. But Trump’s Gulf deal throws that logic into question. Without publicly revising the policy, the U.S. appears to be treating the UAE and Saudi Arabia as Tier 1 actors. This weakens the coherence of the original framework. and may invite other countries to lobby for similar treatment.
Critics warn that this is a recipe for leaks. What if AI chips or software end up in Chinese hands through intermediaries? What if Gulf-built models are reverse engineered, copied, or repurposed? But that may miss the bigger picture.
This isn’t about containment; it’s also about alignment. Washington may see this as a hedge: better to embed the Gulf in U.S.-centered platforms than to let them drift into Beijing’s tech ecosystem. A world of perfect control may be gone. In its place is a new model: managed exposure, with political and technological strings attached. This is the logic of alliances: we share risk, not because it’s completely safe, but because it’s strategic.
Final Thoughts
Trump’s Gulf chip deal isn’t just a transaction. It’s a test case for a new approach to tech leadership. In a world where absolute control is slipping, the U.S. is learning to build bridges instead of walls, sharing key technologies in exchange for influence, trust, and embedded infrastructure.
The next decade of AI will not be won by hoarding chips or restricting data. It will be won by shaping the rules, flows, and alliances that determine how those chips and data are used. If America can manage that balance, strategic sharing without strategic surrender, it may stay on top. If not, someone else will.