Nvidia GPUs matter to Saudi Arabia because compute access is now a bottleneck for national AI strategy. Saudi Arabia can fund data centers, train engineers, and create companies such as HUMAIN, but frontier AI still depends on scarce accelerators, high-speed networking, export approvals, power, cooling, and trusted operations. The nvidia saudi partnership is therefore not just a hardware procurement story. It is a test of whether Saudi sovereign AI infrastructure can scale inside US export-control rules, supplier politics, and Vision 2030 delivery constraints. Commerce has authorized specific HUMAIN purchases under security and reporting conditions, but that is not unrestricted access and it is not proof that every announced GPU is already deployed [S7].
The useful way to read searches for “nvidia saudi arabia,” “nvidia humain,” “humain nvidia,” “nvidia ai chips saudi arabia,” or “nvidia ai chips sale saudi arabia” is as one question: how much frontier compute can Saudi Arabia actually obtain, operate, and govern? The answer is more serious than the headline chip counts. NVIDIA announced a first phase of 18,000 GB300 Grace Blackwell systems with HUMAIN and a projected Saudi AI-factory buildout of up to 500 megawatts over five years, powered by several hundred thousand NVIDIA GPUs [S8]. Commerce later said HUMAIN had authorization to purchase the equivalent of up to 35,000 NVIDIA GB300-equivalent chips under conditions [S7]. Those two facts are compatible, but they are not identical.
This article is the compute-and-export-control node in the wider Saudi AI map. For the broader company profile, see the HUMAIN AI strategy briefing. For the public-governance layer, see the SDAIA analysis. For sector context, see Saudi AI strategy and Saudi data centers.
Why Nvidia GPUs became strategic infrastructure for Saudi AI
Compute is the bottleneck behind Saudi sovereign AI
Saudi Arabia’s AI ambition is not limited to chatbots or pilot projects. Vision 2030 needs AI capacity for government services, energy operations, logistics, healthcare, tourism, Arabic-language models, industrial optimization, cyber defense, and private-sector productivity. Those use cases all depend on infrastructure: data, models, cloud platforms, chips, software frameworks, power, cooling, and operational reliability.
The central constraint is compute. Frontier training and high-volume inference are not possible at national scale without large accelerator clusters. A Saudi gpu data center is therefore a strategic asset, not merely a server room. It controls who can train models, where sensitive data can be processed, which companies can buy AI services locally, and whether Saudi Arabia remains a consumer of imported AI or becomes a regional supplier of compute.
PIF launched HUMAIN on May 12, 2025 as a PIF-owned AI company chaired by Crown Prince Mohammed bin Salman and positioned across next-generation data centers, AI infrastructure, cloud capabilities, advanced models, and AI solutions [S11]. That made HUMAIN a national platform rather than a narrow software venture. Its role overlaps with the PIF capital-allocation model and with Saudi Arabia’s broader attempt to use sovereign balance-sheet strength to build industrial capacity.
The logic is simple but difficult to execute. If Saudi Arabia can secure controlled access to NVIDIA GPUs and operate them in compliant, high-density facilities, it can build sovereign AI infrastructure and attract global customers. If access is delayed, restricted, or politically unstable, Saudi AI strategy has to rely more heavily on lower-tier compute, partner clouds, or non-US suppliers.
Why Blackwell, InfiniBand, and GPU clusters matter
The specific chip generation matters because AI infrastructure is an integrated system. NVIDIA’s Saudi announcement referred to GB300 Grace Blackwell systems and NVIDIA networking, including InfiniBand, as part of the Saudi AI-factory architecture [S8]. In a frontier AI cluster, the GPUs do not operate in isolation. They need high-bandwidth memory, interconnects, networking, software libraries, storage, power delivery, cooling, security controls, and operators who can keep utilization high.
That is why “GB300 Saudi” and “Saudi Arabia NVIDIA Blackwell” are not just product keywords. They describe the level of performance Saudi buyers are trying to access. Blackwell-class systems can support more advanced training and inference workloads than older accelerators, but they also trigger more intense export-control scrutiny because they represent strategic computing capability [S1], [S3].
NVIDIA’s own investor disclosures reinforce that data-center GPUs now sit inside geopolitical risk. The company reported that Data Center was the largest revenue driver in fiscal 2026, while warning that export controls and geopolitics can restrict sales of advanced products and affect revenue opportunities in some markets [S9], [S10]. For “nvda saudi arabia” or “nvidia stock saudi arabia” searches, the right framing is not a buy-or-sell view. Saudi demand is an investor-relevant signal of global AI infrastructure demand, but delivery depends on export approvals, customer execution, supply availability, and compliance.
What Nvidia, HUMAIN, and SDAIA actually announced
HUMAIN’s 500 MW AI factory plan
NVIDIA announced on May 13, 2025 that Saudi Arabia and NVIDIA would build AI factories to power new AI capacity, with HUMAIN projected to build up to 500 megawatts of AI-factory capacity over five years using several hundred thousand NVIDIA GPUs [S8]. That is the broadest public basis for the nvidia saudi partnership.
The status language matters. NVIDIA announced a projected buildout. It did not prove that several hundred thousand GPUs were already installed, energized, and serving customers. A five-year AI-factory plan requires export licenses or license-exception eligibility, site selection, power agreements, grid integration, cooling systems, physical security, cyber controls, customer demand, and operational staff.
This is also why the HUMAIN page and this page should not compete. HUMAIN is the company-level vehicle; this article owns the chip-access and export-control question. The better internal path is to treat HUMAIN’s infrastructure strategy as the platform context and this article as the NVIDIA, GB300, and US-controls explainer.
The 18,000 GB300 first phase
The most concrete NVIDIA-HUMAIN chip number in the May 2025 announcement is the first phase: 18,000 NVIDIA GB300 Grace Blackwell systems for a Saudi AI supercomputer [S8]. That is the safest answer to the narrow “HUMAIN GB300” or “KSA NVIDIA chips” question when the user wants a named first phase.
Commerce’s later authorization introduces another number. On November 19, 2025, Commerce said the United States had authorized specific exports to G42 and HUMAIN, allowing each to purchase the equivalent of up to 35,000 NVIDIA GB300-equivalent chips, with security and reporting requirements and BIS compliance monitoring [S7]. This supports a narrower statement: the US authorized a specific HUMAIN purchase ceiling under conditions. It does not support saying Saudi Arabia has unrestricted access to all Blackwell GPUs, or that the full quantity has already been delivered.
The resulting answer to “how many NVIDIA chips is Saudi Arabia buying” is layered. NVIDIA announced a first HUMAIN phase of 18,000 GB300 systems [S8]. NVIDIA also announced a projected five-year HUMAIN buildout powered by several hundred thousand GPUs [S8]. Commerce later authorized HUMAIN purchases equivalent to up to 35,000 GB300-class chips under conditions [S7]. Those figures describe different scopes, dates, and statuses.
SDAIA’s separate sovereign AI factory role
SDAIA is a separate part of the Saudi AI architecture. NVIDIA’s May 2025 release said it and the Saudi Data and AI Authority would deploy up to 5,000 Blackwell GPUs for a sovereign AI factory [S8]. That should be phrased as “would deploy up to” or “announced plans for,” not as a completed possession claim.
The distinction is institutionally important. HUMAIN is a PIF-owned AI company focused on infrastructure, cloud, models, and solutions [S11], [S12]. SDAIA is the national data and AI authority and sits closer to public-sector data, governance, and government AI adoption. A broad SDAIA profile belongs separately; the detailed governance layer is covered in the SDAIA page.
Together, HUMAIN and SDAIA show why Saudi sovereign AI infrastructure is not a single project. It is a stack: sovereign capital, commercial AI factories, public-sector data governance, cloud platforms, Arabic models, cyber requirements, and diplomatic approvals. NVIDIA is a critical supplier in that stack, but not the only institution that determines whether the strategy works.
Why the deal depends on US export controls
The AI Diffusion rule and the non-enforcement problem
U.S. controls remain legally complex. BIS announced a non-enforcement policy for the January 2025 AI Diffusion Rule and planned a formal rescission/replacement, but GAO later concluded the non-enforcement announcement itself is a CRA-covered rule, while the planned rescission was not final. The current eCFR still contains the AIA and advanced-computing control text. Separately, Commerce has authorized specific Saudi/UAE chip exports under security and reporting conditions [S1], [S4], [S5], [S7].
That paragraph is the core guardrail for any serious article about NVIDIA Saudi Arabia export controls. The January 2025 AI Diffusion Rule revised the Export Administration Regulations for advanced computing integrated circuits and certain advanced AI model weights [S1]. BIS then announced on May 13, 2025 that it would rescind the Biden-era AI Diffusion Rule and instructed enforcement officials not to enforce it before compliance obligations took effect [S4]. GAO later held that the BIS non-enforcement announcement was itself a rule covered by the Congressional Review Act, while the planned rescission was not final agency action [S5].
The practical result is neither simple open access nor simple prohibition. The rule text, the non-enforcement posture, the GAO legal decision, and later Commerce authorizations all matter. A Saudi buyer may be strategically aligned with the United States and still need item-specific, end-user-specific, end-use-specific, and condition-specific approvals. OIRA’s Spring 2025 regulatory agenda separately listed the rescinding action at final-rule stage, adding process context but not by itself completing the legal change [S6].
What the Commerce approval for HUMAIN does and does not mean
Commerce’s November 2025 statement is the key official source for the specific Saudi chip approval. It said the United States had authorized exports for G42 in the UAE and HUMAIN in Saudi Arabia, allowing purchases equivalent to up to 35,000 NVIDIA GB300-equivalent chips, contingent on security and reporting requirements [S7]. Commerce also linked the approvals to compliance monitoring [S7].
That supports careful language: “Commerce authorized specific HUMAIN purchases.” It does not support “the US approved all NVIDIA chips for Saudi Arabia.” It also does not prove shipment, installation, acceptance testing, customer availability, or cluster utilization. For a project of this scale, every stage matters: license conditions, procurement, manufacturing allocation, export shipment, data-center readiness, power-on, networking, workload onboarding, and compliance audits.
The authorization is still strategically important. It signals that Washington is willing to permit Saudi access to some advanced AI chips when security conditions are met. It also sets a template for other Gulf AI projects: access to frontier compute may be available, but only inside a monitored framework that addresses diversion, reporting, and security concerns.
Why Saudi Arabia is a partner but not an unrestricted destination
Saudi Arabia should not be described as an adversary in this context. It is a US partner with major investment, defense, energy, and technology ties, and the White House framed the November 2025 US-Saudi relationship around economic, defense, AI, and technology-protection cooperation [S18]. But partner status is not the same as unrestricted AI-chip access.
The current eCFR text for 15 CFR Part 740 contains the AIA license-exception structure and country list, and Saudi Arabia is not listed in Supplement No. 5(a) AIA destinations [S2]. Part 742 continues to contain advanced-computing license requirements for specified items [S3]. That means the compliance analysis depends on classification, destination, end user, end use, license exceptions, individual licenses, and conditions.
For Saudi AI chips, the political reality and the legal reality have to be read together. Washington wants strategic technology partners outside China, but it also wants controls that prevent diversion, uncontrolled re-export, and uncontrolled access by parties of concern. Riyadh wants frontier compute, but it must build the trust, controls, and reporting systems that make sustained access possible.
How the Nvidia Saudi partnership fits Vision 2030
HUMAIN, PIF, SDAIA, cloud, and data centers
The NVIDIA-HUMAIN relationship fits Vision 2030 because compute is becoming industrial infrastructure. A country that can host large AI factories can support local models, cloud services, data residency, enterprise AI, and cross-border compute demand. That is why the NVIDIA announcement sits beside Saudi data-center and cloud initiatives rather than outside them [S8], [S11], [S14].
HUMAIN’s official PIF description emphasizes full-stack AI scope, including data centers, infrastructure, cloud capabilities, advanced models, and solutions [S11], [S12]. PIF and Aramco later announced a non-binding term sheet for Aramco to acquire a significant minority stake in HUMAIN, with PIF retaining majority ownership if the transaction closes [S13]. That potential Aramco role matters because industrial AI demand, energy expertise, and power planning are all relevant to high-density compute.
For readers tracking “Saudi Vision 2030 AI data centers,” the important point is that the NVIDIA relationship is not a standalone purchase order. It connects to Saudi AI cloud and compute infrastructure, the Saudi AI strategy and government adoption stack, PIF ownership, SDAIA governance, and data-center capacity.
AWS, AMD, Qualcomm, xAI, and supplier diversification
Saudi Arabia is not building its AI stack around NVIDIA alone. AWS and HUMAIN announced a more than $5 billion AI Zone in Saudi Arabia, combining AWS services and infrastructure, with AWS also noting a Saudi Region planned for 2026 [S14]. AMD, Cisco, and HUMAIN announced plans for a joint venture expected to begin operations in 2026, with a first phase of 100 megawatts and a target of up to 1 gigawatt by 2030 using AMD Instinct MI450 systems and Cisco infrastructure [S15]. Qualcomm and HUMAIN announced an MOU covering AI data centers, hybrid cloud-to-edge services, and a Saudi semiconductor design-center collaboration with MCIT [S16].
xAI adds a different kind of signal. In November 2025, xAI announced a framework agreement with Saudi Arabia and HUMAIN to design, build, and operate hyperscale GPU data centers and deploy Grok nationwide [S17]. That should be treated as a framework agreement, not a completed Saudi GPU buildout. It still matters because it points to the demand side: Saudi Arabia is not only trying to own compute; it is trying to attract frontier AI workloads and applications.
Supplier diversification is a rational hedge. NVIDIA Blackwell systems are central to the story, but AMD accelerators, AWS infrastructure, Qualcomm inference and edge designs, xAI workloads, and local data-center operators can reduce dependence on any single supplier. Diversification also strengthens negotiating leverage in a market where advanced accelerators are scarce and politically sensitive.
Saudi-US technology diplomacy
The nvidia saudi partnership sits inside broader Saudi-US technology diplomacy. The White House November 2025 fact sheet described an AI memorandum of understanding and technology-protection language as part of the economic and defense partnership [S18]. That matters because the export-control issue is not only a Commerce licensing question. It is also a diplomatic bargain about where American AI technology can scale without weakening US security goals.
The US-Saudi investment and technology deals context is therefore directly relevant. Saudi Arabia wants advanced chips, cloud partnerships, defense continuity, and investment credibility. The United States wants exports, strategic alignment, security assurances, and influence over how frontier AI infrastructure develops outside its own borders.
The bargain can work only if the Saudi side turns announcements into compliant operations. That means documented controls, audited facilities, cyber resilience, end-use monitoring, trusted cloud procedures, and credible separation from restricted actors. Without those foundations, the political case for broader chip access weakens.
Risks that can still slow the strategy
Compliance, reporting, and diversion risk
The first risk is compliance. Export-controlled chips carry conditions, and those conditions can change. Commerce’s authorization for HUMAIN was tied to security and reporting requirements, not open-ended access [S7]. NVIDIA’s own Form 10-K warns that export-control measures can restrict sales of data-center products and create geopolitical, compliance, and revenue risks [S10].
Diversion risk is the reason approvals are conditional. US officials care not only about where a chip is shipped, but who can access the system, which software runs on it, which customers use it, whether data flows are controlled, whether re-export can occur, and whether parties of concern can benefit indirectly. Those concerns are heightened for high-performance clusters because the capability is strategic.
Saudi Arabia can reduce this risk, but not wish it away. The path is boring and operational: clear ownership, vetted operators, strong physical security, auditable logs, cybersecurity controls, customer screening, procurement transparency, and compliance staff who can satisfy US regulators over time.
Power, cooling, water, and talent constraints
The second risk is infrastructure. A 500 MW AI-factory plan is not only a chip-purchase plan [S8]. It requires power availability, grid planning, substations, backup systems, cooling design, water strategy, construction sequencing, networking, land, and skilled operators. High-density AI clusters turn utilities into strategic constraints.
Saudi Arabia has advantages: energy resources, capital, land, a central mandate, and an ambition to localize AI infrastructure. But those advantages do not remove engineering bottlenecks. Power has to arrive at the right site, cooling has to work in a hot climate, water use has to be managed, and enough engineers have to be trained to run large GPU clusters at high utilization.
Talent is especially important. A Saudi sovereign AI infrastructure strategy cannot succeed if the most strategic work remains permanently outsourced. Vision 2030 value comes from operating capability: data-center engineering, chip-cluster operations, AI platform management, model training, inference optimization, cyber defense, and enterprise adoption. Otherwise, Saudi Arabia risks buying expensive capacity without building durable local advantage.
NVDA investor signal versus delivery risk
For NVIDIA, Saudi demand is a sign of how far AI infrastructure demand has globalized. The company reported very large Data Center revenue in fiscal 2026, and Saudi Arabia’s AI-factory announcements fit the pattern of sovereign and hyperscale buyers competing for frontier compute [S9]. That is why “NVDA Saudi Arabia” and “NVIDIA stock Saudi Arabia” searches appear around these announcements.
But investor relevance is not the same as investment advice. The Saudi opportunity is execution-sensitive. Revenue recognition, shipment timing, license conditions, customer funding, facility readiness, and supply allocation all matter. Export controls can slow or reshape data-center GPU sales even when demand exists [S10].
For Saudi Arabia, the same caution applies in reverse. A press release can signal strategy, but the real test is commissioned capacity, compliant operations, customer adoption, and model output. The nvidia saudi partnership is a major signal that Saudi AI has moved from strategy slides toward hard infrastructure. It is not yet proof that Saudi Arabia has solved the compute bottleneck.
FAQ
How many Nvidia chips is Saudi Arabia buying?
There is no single clean number. NVIDIA announced a HUMAIN first phase involving 18,000 GB300 Grace Blackwell systems and a projected five-year buildout of up to 500 MW powered by several hundred thousand NVIDIA GPUs [S8]. NVIDIA also said it and SDAIA would deploy up to 5,000 Blackwell GPUs for a sovereign AI factory [S8]. Separately, Commerce authorized HUMAIN to purchase the equivalent of up to 35,000 GB300-equivalent chips under security and reporting conditions [S7].
Did the US approve Nvidia chips for Saudi Arabia?
Yes, but only in a specific, conditioned sense. Commerce announced on November 19, 2025 that HUMAIN and G42 were authorized to purchase the equivalent of up to 35,000 NVIDIA GB300-equivalent chips, subject to security and reporting requirements [S7]. That does not mean Saudi Arabia has unrestricted Blackwell access or that all authorized chips are already deployed.
What is the Nvidia Saudi partnership?
The Nvidia Saudi partnership refers mainly to NVIDIA’s May 2025 announcement with HUMAIN to build Saudi AI factories, including an 18,000 GB300 first phase and a projected 500 MW five-year buildout, plus NVIDIA’s separate SDAIA sovereign AI factory role [S8]. It is best understood as a compute-infrastructure partnership shaped by export controls.
Is HUMAIN buying Nvidia GB300 chips?
NVIDIA announced an initial HUMAIN phase involving 18,000 GB300 Grace Blackwell systems [S8]. Commerce later authorized HUMAIN to purchase up to a stated GB300-equivalent amount under conditions [S7]. The safe language is “announced,” “authorized,” and “subject to conditions,” not “fully delivered.”
Why are export controls involved?
Advanced AI chips are controlled because they can support strategically significant computing capability. The January 2025 AI Diffusion Rule, the BIS non-enforcement announcement, GAO’s CRA decision, current eCFR text, and Commerce’s later HUMAIN authorization all show that Saudi AI-chip access is governed through a complex policy framework [S1], [S2], [S3], [S4], [S5], [S7].
Does AWS’s HUMAIN AI Zone mean Nvidia chips are guaranteed?
No. AWS and HUMAIN announced a more than $5 billion AI Zone in Saudi Arabia, including AWS services and infrastructure, but that announcement should not be treated as a separate guarantee of NVIDIA chip access [S14]. NVIDIA-specific chip claims should be tied to NVIDIA and Commerce sources [S7], [S8].
Is Saudi Arabia trying to replace Nvidia with AMD or Qualcomm?
No. The better interpretation is supplier diversification. NVIDIA remains central to the announced GB300 and Blackwell strategy [S8]. AMD, Cisco, AWS, Qualcomm, and xAI add alternative compute, cloud, networking, inference, and workload pathways [S14], [S15], [S16], [S17].
Is this good for Nvidia stock?
Saudi AI demand is an investor-relevant signal because it shows sovereign buyers competing for frontier data-center capacity [S9]. But it is not a buy or sell recommendation. Export controls, licensing, delivery timing, facility readiness, and geopolitical risk can affect whether announced demand turns into recognized revenue [S10].
Sources
- [S1] GovInfo, 90 FR 4544, Framework for Artificial Intelligence Diffusion — official Federal Register rule, Jan. 15, 2025. https://www.govinfo.gov/app/details/FR-2025-01-15/2025-00636
- [S2] eCFR, 15 CFR Part 740 — current official CFR/unofficial edition, current to May 28, 2026. https://www.ecfr.gov/current/title-15/subtitle-B/chapter-VII/subchapter-C/part-740
- [S3] eCFR, 15 CFR Part 742 — current official CFR/unofficial edition, current to May 28, 2026. https://www.ecfr.gov/current/title-15/subtitle-B/chapter-VII/subchapter-C/part-742
- [S4] BIS, Department of Commerce Announces Rescission of Biden-Era Artificial Intelligence Diffusion Rule — official agency statement, May 13, 2025. https://www.bis.gov/press-release/department-commerce-announces-rescission-biden-era-artificial-intelligence-diffusion-rule-strengthens
- [S5] GAO B-337935, Department of Commerce, Bureau of Industry and Security: Rescission of a Rule — institutional legal decision, May 12, 2026. https://www.gao.gov/products/b-337935
- [S6] OIRA, RIN 0694-AJ90, Rescinding the Framework for Artificial Intelligence Diffusion — official regulatory agenda, Spring 2025. https://mobile.reginfo.gov/public/do/eAgendaViewRule?RIN=0694-AJ90&pubId=202504
- [S7] U.S. Department of Commerce, Statement on UAE and Saudi Chip Exports — official agency statement, Nov. 19, 2025. https://www.commerce.gov/news/press-releases/2025/11/statement-uae-and-saudi-chip-exports
- [S8] NVIDIA, Saudi Arabia and NVIDIA to Build AI Factories to Power Next Wave of Intelligence for the Age of Reasoning — company primary, May 13, 2025. https://nvidianews.nvidia.com/news/saudi-arabia-and-nvidia-to-build-ai-factories-to-power-next-wave-of-intelligence-for-the-age-of-reasoning
- [S9] NVIDIA, Announces Financial Results for Fourth Quarter and Fiscal 2026 — company primary/investor, Feb. 25, 2026. https://investor.nvidia.com/news/press-release-details/2026/NVIDIA-Announces-Financial-Results-for-Fourth-Quarter-and-Fiscal-2026/
- [S10] NVIDIA, FY2026 Form 10-K — SEC filing, Feb. 25, 2026. https://www.sec.gov/Archives/edgar/data/1045810/000104581026000021/nvda-20260125.htm
- [S11] PIF, HRH Crown Prince launches HUMAIN as global AI powerhouse — Saudi official/primary, May 12, 2025. https://www.pif.gov.sa/en/news-and-insights/press-releases/2025/hrh-crown-prince-launches-humain-as-global-ai-powerhouse/
- [S12] PIF, HUMAIN portfolio page — Saudi official/primary, current. https://www.pif.gov.sa/en/our-investments/our-portfolio/humain/
- [S13] PIF, PIF and Aramco agree for Aramco to acquire a significant minority stake in HUMAIN with PIF retaining majority ownership — Saudi official/primary, Oct. 28, 2025. https://www.pif.gov.sa/en/news-and-insights/press-releases/2025/pif-and-aramco-agree-for-aramco-to-acquire-a-significant-minority-stake-in-humain-with-pif-retaining-majority-ownership/
- [S14] AWS, AWS and HUMAIN announce groundbreaking AI Zone to accelerate AI adoption in Saudi Arabia and globally — company primary, May 13, 2025. https://press.aboutamazon.com/2025/5/aws-and-humain-announce-groundbreaking-ai-zone-to-accelerate-ai-adoption-in-saudi-arabia-and-globally
- [S15] AMD, Cisco and HUMAIN to form joint venture — company primary, Nov. 19, 2025. https://www.amd.com/en/newsroom/press-releases/2025-11-19-amd-cisco-and-humain-to-form-joint-venture.html
- [S16] Qualcomm, Qualcomm and HUMAIN to develop state-of-the-art AI data centers — company primary, May 13, 2025. https://www.qualcomm.com/news/releases/2025/05/qualcomm-and-humain-to-develop-state-of-the-art-ai-data-centers-
- [S17] xAI, Grok Goes Global — company primary, Nov. 19, 2025. https://x.ai/news/grok-goes-global/
- [S18] White House, President Donald J. Trump Solidifies Economic and Defense Partnership with the Kingdom of Saudi Arabia — official executive branch fact sheet, Nov. 18, 2025. https://www.whitehouse.gov/fact-sheets/2025/11/fact-sheet-president-donald-j-trump-solidifies-economic-and-defense-partnership-with-the-kingdom-of-saudi-arabia/