PetroCompute: Will A.I.'s Future Run Through the Gulf?

Power constraints and geopolitical rivalry position the G.C.C. to be an A.I. inference powerhouse

Operators feed stories into an IBM computer programmed in Arabic and connected to an electronic Linotype machine, capable of setting 1,000 lines per hour. Image credit: © Aramco World / Nik Wheeler / 1972

G.C.C. at the Crossroads

Kuwait's fortunes underwent two profound transformations in the early 20th century. Positioned at the northern tip of the Gulf, the town had established a thriving economy built on pearling and shipbuilding, activating extensive trade networks with India and East Africa. When Japanese cultured pearls and British steamships decimated these traditional industries in the 1920s, the discovery of commercial oil reserves proved transformative.

This 1938 discovery coincided with two global shifts: the great powers’ military transition from coal to oil-powered fleets and the emergence of mass automobile production. This convergence transformed Kuwait and its Gulf neighbours from mercantile townships into global energy players and, in due course, capital allocators with their natural resource endowment defining the century that followed. Today, the oil-producing Gulf Cooperation Council (G.C.C.) states confront a similar inflection point, though one demanding strategic foresight rather than serendipity.

The G.C.C.'s current economic model faces mounting pressures. The hydrocarbon landscape has fundamentally shifted since the 2014 oil price crash, shaped by U.S. shale production, weakening global demand, and persistent market oversupply. The region's position, however, remains more robust than conventional wisdom suggests: G.C.C. states maintain their status as the world's most efficient oil producers, with production costs below $10 per barrel - a stark contrast to U.S. shale oil costs exceeding $50.¹ ²

While low production costs may secure greater market share even in a contracting oil market, this advantage masks future structural risks. As global energy transition accelerates, the G.C.C. states' growing share of a contracting market amplifies their fiscal exposure to oil price volatility and weakens their strategic influence, which has historically stemmed from their role in global energy security. This deepening dependence threatens both fiscal sustainability and, more fundamentally, the G.C.C.'s position within global power structures that have underpinned regional stability for decades.

Compute as the New Currency for Power

Artificial intelligence’s latest ascent presents the G.C.C. states with a compelling opportunity. The emergence of generative A.I. models have sparked a global race for computational supremacy, as breakthroughs in the technology enable increasingly powerful applications from text and video generation to intelligent autonomous agents. This competition centers on three critical elements: data and algorithms to create A.I. models, advanced semiconductors to train these models, and - most crucially for the G.C.C. - the power-intensive computing infrastructure needed for inference, the process of deploying trained A.I. models.

Global investment in A.I. infrastructure - from semiconductors to data centres and power generation - has reached extraordinary scale. U.S. technology giants plan to invest $300 billion in 2025 alone, more than double U.S. oil and gas industry spending.³ ⁴ Nations from India to China are mobilizing similarly, while McKinsey projects infrastructure investment will exceed $1 trillion⁵ within a few years, drawing large commitments from public and private investors alike.⁶ For both companies and nations, this race mirrors Pascal's Wager: the cost of non-participation could exceed the investment itself.

Nowhere does this wager play out more prominently than in the U.S.-China rivalry: China's $150 billion investment in semiconductor development over the past decade prompted a decisive U.S. response through the CHIPS and Science Act of 2022.⁷ The legislation allocated $53 billion for semiconductor manufacturing and research initiatives and catalyzed over $640 billion in private investment, underscoring Washington's determination to prevent China from achieving technological parity - a scenario it considers a critical national security threat.⁸

The G.C.C. has emerged as a key arena of U.S.-China geoeconomic competition over the past decade, echoing the position of Japan, South Korea, and Taiwan in the decades leading up to their economic ascendance. These Asian economies provide a blueprint more than a historical parallel: they leveraged U.S. Cold War strategic interests at the time to access emerging technology and build exceptional sovereign capabilities. Their success demonstrates how early positioning in critical infrastructure, combined with strategic U.S. alignment, can create lasting economic and security advantages.

Early Positioning

As rising middle powers, the G.C.C. states - particularly Saudi Arabia and the U.A.E. - are already positioning themselves as emerging players within this global race for A.I. infrastructure.

Initially, Saudi Arabia and the U.A.E. pursued a balanced approach, maintaining hardware purchases, investments, and research collaborations with both the U.S. and China.⁹ This strategy prompted U.S. export restrictions on advanced semiconductors to all G.C.C. countries in October 2023, driven by concerns over technology transfer to China.¹⁰ However, it also opened channels for dialogue with U.S. policymakers increasingly concerned about China's growing regional influence.¹¹

In April 2024, Microsoft invested $1.5 billion in G42, a U.A.E. government-backed A.I. firm, to jointly deploy A.I. infrastructure across the Middle East, Central Asia, and Africa. Beyond capital, the deal marks a crucial realignment between the U.S. and G.C.C. states and establishes a preliminary framework for A.I. collaboration. The U.S. Commerce Department’s Bureau of Industry and Security (B.I.S.) required G42 to sever ties with Chinese companies like ByteDance and Huawei before approving the Microsoft investment.¹² This move echoes U.S. technology transfer policies that helped forge semiconductor industries in Japan and subsequently in South Korea and Taiwan — creating a strategic buffer against Soviet and, later, Chinese influence. Similar dynamics could enable Gulf states to secure vital trade, security, and technology agreements amid U.S.-China tensions - effectively strengthening their regional posture and securing their place on the proverbial A.I. table.

Several G.C.C. states have made initial moves to establish regional A.I. leadership, with the U.A.E. emerging as the pacesetter. Its sovereign wealth fund Mubadala controls GlobalFoundries - which produces 6 percent of global semiconductors - while also holding stakes in frontier A.I. firms like Anthropic. Two Emirati A.I. focused vehicles, G42 and MGX, are spearheading domestic and regional A.I. infrastructure development and securing strategic stakes across the A.I. value chain - from investments in OpenAI and supercomputer manufacturer Cerebras Systems to development of a highly ranked open-source large language model. A $30 billion A.I. infrastructure fund partnership between MGX, Microsoft, and BlackRock underscores these ambitions. Yet these technological alliances remain delicate: Cerebras Systems' planned public offering has encountered headwinds following U.S. national security scrutiny of G42's minority stake.¹³

Saudi Arabia has pursued a more targeted yet ambitious approach to A.I. The Kingdom's planned $100 billion A.I. infrastructure initiative, launched through major deals with Google Cloud and inference chipmaker Groq, underscores its ambition to make technology a cornerstone of its economic transformation.¹⁴

Washington’s behind-the-scenes involvement in these deals resembles its strategic playbook from the 1980s, when it cultivated semiconductor industries in South Korea and Taiwan to counter Japan's memory chip dominance. The parallel suggests the U.S. may, through technology companies, strategically distribute support for A.I. infrastructure and capabilities across G.C.C. states and therefore avoiding over-reliance on a single regional partner while maintaining influence throughout the region.

Strategic Opening

These initiatives only hint at the G.C.C. states' fundamental opportunity: building high performance compute infrastructure of unprecedented scale to serve the world's rapidly growing demand for A.I.

The magnitude of this undertaking becomes clear through A.I.'s escalating power requirements. A decade ago, a 30-megawatt (MW) data centre was considered substantial; today, the largest A.I. model training facilities require up to 300 MW, while next-generation sites are being planned at scales of 1.0 to 5.0 GW - sufficient to power nearly four million homes.¹⁵ ¹⁶

This computational growth coincides with looming power constraints in the U.S. A recent Institute for Progress study projects that while global A.I. data centers will require an additional 130 GW of power by 2030, U.S. gas-power generation capacity will increase by only 30 GW - most unavailable for A.I. data centres. While there is expected to be an incremental 15 GW of renewable capacity being added to the U.S. grid annually, such supply is intermittent and not ideally suited for the uninterruptible power demands from A.I. data centres. These domestic constraints may force U.S. policymakers to prioritize power capacity for strategically vital A.I. model training while moving inference operations, the less critical task of running these models, abroad.¹⁷

This convergence of A.I.'s escalating power requirements with U.S. infrastructure constraints opens a strategic path for G.C.C. states to become major suppliers of A.I. inference through gigawatt-scale computing infrastructure. Unlike training, inference computing requires less advanced semiconductors - which face fewer export restrictions - enabling the G.C.C. to build substantial capacity even under current U.S. export controls.¹⁸

While training A.I. models demands massive computational resources, the cumulative power needs from billions, and eventually trillions, of daily inference operations will exceed training requirements. This disparity widens as A.I. model performance improvements increasingly depend on inference-time reasoning - where models perform complex calculations while serving users rather than at training - and as models tackle more power-intensive tasks like video generation and continuous autonomous operations by A.I. agents. To quantify this trajectory: a single ChatGPT query already requires ten times the processing power of a Google search, while a generative A.I. video demands 10,000 times more.¹⁹

The Triple Advantage

Beyond the U.S. and China, the G.C.C. states possess three structural advantages that reinforce and amplify each other, creating unique capabilities to serve global A.I. inference demand.

Most critically, G.C.C. states’ advantage in power infrastructure - low energy costs, relatively modern generation and transmission assets, and ability to rapidly deploy new capacity - position them to rapidly scale A.I. infrastructure. Unsubsidised power costs in G.C.C. states average $0.10²⁰ per kilowatt-hour (kWh), compared to $0.17²¹ in the U.S. and $0.29²² in Europe. Through centralized planning and execution, G.C.C. states deploy power infrastructure at remarkable scale and speed: Saudi Arabia plans to add 42 GW of gas combined-cycle power capacity by 2030²³ - 40 percent more than the entire U.S. during the same period.

This energy advantage is complemented by strategic geography: the G.C.C.'s position at the crossroads of three continents, combined with its extensive submarine cable connectivity, enables it to serve four billion internet users within 100-millisecond latency²⁴ - the threshold for perceiving A.I. responses as instantaneous. This positions the Gulf states to serve the majority of global inference workloads. Moreover, the G.C.C.’s extensive coastline and arid climate have necessitated the world's largest desalination infrastructure, producing 40 percent of global desalinated water output²⁵ - an established capability that provides needed water cooling for power-intensive A.I. data centres as the infrastructure scales to unprecedented levels.

Financial capability represents a third decisive advantage: the G.C.C. states have substantial liquid capital endowment as evidenced by nearly $5 trillion²⁶ in sovereign wealth fund assets accumulated through decades of compounding investments across U.S., European, and Asian markets. The G.C.C. states combine a long-term investment horizon with proven capital deployment capability - demonstrated through $320 billion in annual capital spending across hydrocarbon infrastructure ($115 billion), military equipment ($130 billion), and strategic projects ($75 billion).²⁷ This rare combination of patient capital and execution agility at large scale, enabled by centralized decision-making and compounding capital endowment, allows this group of states to pursue strategic infrastructure at a scale and speed that few other countries and regions can match.

Navigating a Sea of Risk

While these structural advantages position G.C.C. states favorably, A.I. infrastructure presents distinct risks requiring careful examination.

First, the G.C.C. could face significant challenges in capturing long-term value from this infrastructure, and therefore value accretion warrants scrutiny. As Malaysia, India, and other nations construct gigawatt-scale A.I. infrastructure, G.C.C. states risk investing in infrastructure that, while necessary, lacks differentiation - a potentially commoditised asset with high fixed costs and limited pricing power. The telecom sector offers a stark warning: when Apple launched the iPhone in 2007, its exclusive launch partner AT&T was valued at $250 billion to Apple's $105 billion. Today, AT&T's market capitalisation stands at $165 billion, while Apple has reached $3.5 trillion - demonstrating how value can accrue to higher layers in the technology stack.²⁸

Early G.C.C. partnerships with U.S. hyperscalers reflect these value capture challenges. Though specific terms are undisclosed, hyperscalers appear to secure both commercial benefits - through a combination of preferential terms on power, land, and tax - and financial advantages, as their direct infrastructure ownership allows them to capitalize rather than expense the cloud business’s growth. Hyperscalers leverage this growing footprint to expand their customer base and capture greater value from the suite of proprietary applications that sits at the top of the A.I. stack. G.C.C. states, in return, receive modest foreign direct investment, jobs, and workforce training - effectively gaining limited strategic and financial value. Future deals must be structured to create enduring value for G.C.C. states, ensuring the returns on the infrastructure translate into lasting strategic and economic benefits.

Geopolitical vulnerability stands as a second risk. Expanding U.S. export controls to less advanced semiconductors threaten to curtail G.C.C. ambitions in A.I. inference infrastructure development. A growing view in Washington of G.C.C. states as 'swing states' between U.S. and Chinese geoeconomic interests complicates matters further. While G.C.C. states have historically aligned with U.S. strategic interests, increasing their technological dependence - mirroring their reliance on U.S. military equipment - could further constrain diplomatic flexibility. The April 2020 oil price collapse illustrated this dynamic: U.S. senators pressured to block military equipment sales to Saudi Arabia unless it cut production to support struggling U.S. shale oil producers - effectively asking the Kingdom to sacrifice market share to benefit direct competitors.²⁹

Beyond geopolitical constraints, the physical infrastructure itself presents a third risk, as technological obsolescence presents a real challenge. Training and deploying new A.I. models demands continuous innovation in semiconductors to handle exponential growth in data processing and model complexity. While traditional semiconductor advancement followed Moore's Law of doubling computational power every 18-24 months, A.I. chip capabilities are now doubling every 4-9 months.³⁰ This acceleration demands not only new chips but also ever-higher power densities and advanced cooling systems, possibly relegating recently built data centres to lower-value use cases. The sustainability of this accelerated pace remains debated within the industry, therefore forecasting future returns and cash flows are subject to considerable uncertainty. The shift from mainframes to distributed computing in the late 1980s offers a cautionary tale of how seemingly indispensable infrastructure can quickly become stranded.³¹

Energy demands pose a fourth risk. In 2023, A.I. data centres globally consumed approximately 12 terawatt-hours (TWh) of electricity³² - equivalent to 2 percent of G.C.C.'s annual power consumption. By 2030, this demand is projected to reach 330 TWh³³, which equates to half the G.C.C.'s total annual power consumption. These conservative estimates exclude emerging A.I. capabilities mentioned previously, which will substantially increase power demands. Even as technological advances reduce power consumption per unit of compute, Jevons Paradox suggests these efficiency gains will paradoxically increase total power use by stimulating greater demand.

Therefore, exponential A.I. inference demand growth could force G.C.C. states to reimagine their power generation strategy - continuing to rely predominantly on hydrocarbons for both domestic and A.I. power needs would create an inevitable ceiling on compute capacity. Serving additional demand would require diverting exportable hydrocarbons to domestic power generation - impacting export revenues and current account balances - while also conflicting with the ambitious climate targets established in recent years. Failure to meet growing A.I. compute demand would risk ceding hard-won market share just when the benefits of scale compound. Power generation and A.I. infrastructure must therefore evolve in tandem – otherwise, the G.C.C. states' ambitions would be constrained by design.

Outsized Dividends

Despite these challenges, A.I. infrastructure development offers G.C.C. states an opportunity to embed themselves in an industry that could transform their economies and build vital complexity.

Ireland's thriving data centre industry demonstrates how early infrastructure positioning can create lasting economic value. Starting with Microsoft's first data center outside the U.S. in 2004, Ireland attracted successive waves of hyperscalers to become the world's third-largest data centre hub, housing 5% of global server capacity. The economic impact has been substantial: the sector directly employs 16,000 people (27,000 including contractors), and global data center leader Equinix alone contributes over $15 billion annually to Ireland's economy - 3% of the country's GDP.³⁴ Beyond direct economic benefits, Ireland's advantages - common law legal system, stable regulations, highly-educated workforce, and concentrated computing infrastructure - have created a powerful network effect, attracting global technology companies to establish hubs in the country.

Just as Ireland positioned itself for the cloud computing era, the G.C.C. states are uniquely placed to capture the A.I. infrastructure wave at a much larger scale. Scale economies from massive A.I. infrastructure presents the first benefit. Leveraging its low-cost energy advantage, the G.C.C. could establish itself as the lowest-cost provider of A.I. compute, mirroring its position in global energy markets. This advantage is particularly relevant for inference computing, where competitive pricing could dictate market share. Such scale and cost advantages would capture two distinct markets: real-time inference demand from neighbouring markets requiring low latency, and price-sensitive workloads for video generation and autonomous agents that can tolerate higher latency. Through disciplined execution and low power costs at scale, the G.C.C. could serve a majority of global inference demand - and training demand if export restrictions ease - cementing its importance within the A.I. value chain.

To capture value from these scale advantages, G.C.C. states would need to move beyond basic infrastructure provision to developing specialised capabilities. Ireland's experience demonstrates a proven path: local contractors that began by building hyperscaler data centres have evolved into preferred global infrastructure partners, supporting hyperscaler data centre buildouts worldwide. This progression from basic construction to specialised services illustrates how G.C.C. states could build competitive advantages beyond low-cost power provision.

The G.C.C. states supporting homegrown A.I. cloud operators to serve hyperscalers and enterprise customers presents an even more compelling opportunity. New industry players like CoreWeave demonstrate the A.I. cloud sector’s potential, achieving 85 percent gross margins with revenues surging from $25 million to $465 million within a year³⁵ - exceeding the already attractive 50 percent margins of traditional data centers. G.C.C. states could participate in the compute-for-equity model by having local A.I. cloud operators provide bulk compute at preferential rates to sovereign wealth funds, who can then leverage it to gain deal access in A.I. companies - following the playbook of Microsoft with OpenAI, Amazon with Anthropic, and venture firms like Andreessen Horowitz. This approach creates a virtuous cycle: increased infrastructure utilization, equity exposure to the application layer where significant value is likely to accrue, and strategic positioning in global A.I. value chain.

Technical capability development presents another benefit. In the 1980s, Taiwan's government established the Industrial Technology Research Institute to facilitate semiconductor technology transfers with U.S. firms like RCA; such industry-academia partnerships supplied the talent pool needed to transform the country from basic manufacturing to advanced semiconductor leadership.³⁶ The G.C.C. states could replicate this approach through hyperscaler partnerships with domestic technical institutes, building the technical capabilities needed to transform local A.I. cloud operators into globally competitive champions.

Realizing these A.I. ambitions will require G.C.C. states to reimagine power generation. While natural gas offers an immediate solution, solar power presents a compelling medium-term option for the region. Located in the global sunbelt, the G.C.C. enjoys exceptional solar radiation levels ideal for power generation. Paired with industrial-scale battery storage, solar power offers a promising path toward delivering blended power capacity for A.I. infrastructure. At less than $0.02 per kilowatt hour (kWh), solar has become the lowest-cost option for power production in the G.C.C., outperforming natural gas, oil, and even nuclear power.³⁷ Saudi Arabia's ambitious plan to build 130 GW of solar capacity over the next five years exemplifies the G.C.C.'s broader push to diversify its power generation sources and uphold its commitment to clean energy transition.³⁸

Nuclear power, with its stable and continuous generation capacity, presents a compelling long-term option for the G.C.C. The U.A.E.'s Barakah facility demonstrates nuclear viability in the region - its 5.6 GW capacity powers 25 percent of the country’s domestic needs, and is projected to deliver electricity at a quarter the cost of that from gas.³⁹ Recent advances in small modular nuclear reactors strengthen this case: these compact units are sized up to 400 MW and can be constructed in under five years with one-tenth the footprint, enabling precise capacity expansion and co-location with A.I. data centres. Introducing nuclear power would further free up hydrocarbons for export that would otherwise fuel domestic power generation, enhance energy security, and provide waste heat for local manufacturing and desalination facilities that G.C.C. states depend on for fresh water.

Therefore, developing gigawatt-scale A.I. infrastructure would establish the G.C.C. as home to one of the world's largest power generation and transmission networks. Historical precedent shows how overbuilding power infrastructure can seed future economic growth: Japan's expansion of power generation in the 1960s enabled its rise as a manufacturing powerhouse, driving annual industrial output growth of 12 percent and annual export growth of 17 percent in the ensuing decade.⁴⁰ Similarly, robust power infrastructure in the G.C.C. could attract energy-intensive industries from automotive manufacturing to green hydrogen production and semiconductor fabrication - provided the necessary technical talent and trade agreements are in place.

These developments collectively point toward the G.C.C.'s ultimate goal: economic diversification through interconnected industries that create and compound technical capabilities. Just as U.S. hyperscalers moved beyond operating data centres to designing custom inference chips - securing their supply chains while achieving superior processing capabilities - G.C.C. A.I. cloud operators could develop upstream capabilities in semiconductor and infrastructure technologies. Meanwhile, building and maintaining this infrastructure would foster downstream industries from component manufacturing to infrastructure construction and maintenance. This comprehensive industrial ecosystem would create the kind of economic complexity that distinguishes advanced economies where knowledge and capabilities accumulate and compound across multiple sectors, thereby reducing G.C.C. states’ dependence on resource extraction.

Charting the Course

For half a century, oil-producing G.C.C. states have stabilized global energy markets through excess capacity and reliable supply. A similar opportunity now emerges in A.I. infrastructure. Success in this domain resembles semiconductor fabrication more than oil production - it is dynamic and relentlessly competitive.

However, the G.C.C. states' path ahead raises vital questions. Will they build the technical capabilities needed to break free from resource dependence and create enduring value, or will their investments simply enable others' ambitions? Most fundamentally, can G.C.C. states leverage their financial resources and geopolitical position to carve out a place in the A.I. value chain, and will great powers permit them to do so?

The G.C.C. states’ future in the global order may depend less on the oil beneath their land than on the computing power they can deploy above it. Their execution amid these uncertainties will determine whether the future of power - geopolitical, electrical, and computational - runs through or past the Gulf.

---

Thanks to Faris Al-Sulayman, Nafez Dakkak, Mashhour Al-Ibrahim, Nikhil Bharadwaj, and Taahir Khamissa for reviewing drafts of this essay. Special thanks to Mohammed Algarawi for engaging with me on screenplay ideas about a futuristic G.C.C. — those conversations helped inspire this essay.

Additional thanks to Perplexity for research assistance and Anthropic’s Claude 3.5 Sonnet for copyediting support.

---

Click here for an A.I. generated podcast episode about this essay.

1. Federal Reserve Bank of Dallas, "Energy Slideshow," Federal Reserve Bank of Dallas, updated 19 September, 2024, https://www.dallasfed.org/-/media/Documents/research/energy/energycharts.pdf

2. Lutz Kilian, "How the Saudi Decision to Launch a Price War Is Reshaping the Global Oil Market," Federal Reserve Bank of Dallas, April 2, 2020, https://www.dallasfed.org/research/economics/2020/0402

3. Eugene Kim, "Big Tech Is Going Wild on AI Spending Next Year and Beyond," Business Insider, November 5, 2024, https://www.businessinsider.com/amazon-microsoft-google-meta-wild-spending-ai-infrastructure-cloud-2024-11

4. Statista Research Department, "U.S. Oil and Gas Industry Capex 2010-2023," Statista, October 4, 2024, https://www.statista.com/statistics/294607/total-us-gas-and-oil-industry-expenditure/

5. McKinsey & Company, "AI Power: Expanding Data Center Capacity to Meet Growing Demand," McKinsey & Company Insights, October 29, 2024, https://www.mckinsey.com/industries/technology-media-and-telecommunications/our-insights/ai-power-expanding-data-center-capacity-to-meet-growing-demand

6. Allison McNeely and Naureen S. Malik, "KKR Sees $250 Billion Spent Annually From Data Center Boom," Bloomberg, November 1, 2024, https://www.bloomberg.com/news/articles/2024-11-01/data-center-boom-seen-fueling-250-billion-of-annual-spending

7. Frank Umbach, "The Geopolitics of the Escalating Chip War," GIS Reports Online, April 29, 2024, https://www.gisreportsonline.com/r/escalating-chip-war/

8. New York Center for Foreign Policy Affairs, "CHIPS Act Sparks $640 Billion Private Investment Surge," New York Center for Foreign Policy Affairs, March 1, 2024, https://nycfpa.org/03/01/2022-chips-act-sparks-640b-private-investment-surge/

9. Eleanor Olcott, "Saudi Fund Invests in China Effort to Create Rival to OpenAI," Financial Times, May 31, 2024, https://www.ft.com/content/87a40ffe-c791-4c90-8123-3f75aa0ed26b

10. Norton Rose Fulbright, "US Expands Export Restrictions on Advanced Semiconductors," Norton Rose Fulbright Publications, October 2023, https://www.nortonrosefulbright.com/en/knowledge/publications/5a936192/us-expands-export-restrictions-on-advanced-semiconductors

11. Mark Bergen, Mackenzie Hawkins, Dina Bass, and Ben Bartenstein, "Microsoft AI Deal With UAE’s G42 at Risk Over National Security Fears," Bloomberg, July 2, 2024, https://www.bloomberg.com/news/articles/2024-07-02/microsoft-ai-deal-with-uae-s-g42-at-risk-over-national-security-fears

12. Paolo Confino, "‘We Cannot Work With Both Sides’: A Major Emirati AI Company Has Picked a Side in the U.S.-China Tech War," Fortune, December 8, 2023, https://fortune.com/2023/12/07/emirati-ai-company-picks-side-against-china-g24-us-sanctions/

13. Echo Wang, Max A. Cherney, and Krystal Hu, "Cerebras Likely to Postpone IPO Due to CFIUS Review Delay of G42 Deal, Sources Say," Reuters, October 8, 2024, https://www.reuters.com/technology/cerebras-likely-postpone-ipo-due-cfius-review-delay-g42-deal-sources-say-2024-10-08/

14. Marissa Newman, Julia Love, Mark Bergen, and Christine Burke, "Saudis Plan $100 Billion AI Powerhouse to Rival UAE’s Tech Hub," Bloomberg, November 6, 2024, https://www.bloomberg.com/news/articles/2024-11-06/saudis-plan-100-billion-ai-powerhouse-to-rival-uae-s-tech-hub

15. Jacqueline Davis, "Large Data Centers Are Mostly More Efficient, Analysis Confirms," Uptime Institute Journal, February 7, 2024, https://journal.uptimeinstitute.com/large-data-centers-are-mostly-more-efficient-analysis-confirms/

16. Ben Cottier, Robi Rahman, Loredana Fattorini, Nestor Maslej, and David Owen, "The Rising Costs of Training Frontier AI Models," arXiv, May 31, 2024, https://arxiv.org/abs/2405.21015

17. Tim Fist and Arnab Datta, "How to Build the Future of AI in the United States," Institute for Progress, October 23, 2024, https://ifp.org/future-of-ai-compute/

18. Saif M. Khan and Alexander Mann, "AI Chips: What They Are and Why They Matter," Center for Security and Emerging Technology, Georgetown University, April 2020, https://cset.georgetown.edu/wp-content/uploads/AI-Chips%E2%80%94What-They-Are-and-Why-They-Matter-1.pdf

19. Sean Klimczak, "The Convergence of Data Centers and Power: A Generational Investment Opportunity," Blackstone, October 31, 2024, https://www.blackstone.com/insights/article/the-convergence-of-data-centers-and-power-a-generational-investment-opportunity-the-connection/

20. Glnar Eskandar, "Missed Opportunities: Billions Sacrificed Annually to Generate Electricity in the GCC," Middle East Institute, October 17, 2023, https://mei.edu/publications/missed-opportunities-billions-sacrificed-annually-generate-electricity-gcc

21. Jamie Smith, Catherine Lane, Deborah Foley, "Compare Electricity Rates and Average Bill Costs in Each State in 2024," SolarReviews, accessed October 30, 2024, https://www.solarreviews.com/blog/average-electric-rate-data

22. Qery, "Consumer Energy Prices in Europe," Qery, accessed October 2024, https://qery.no/consumer-energy-prices-in-europe

23. H.R.H Prince Abdulaziz bin Salman Al Saud, "KSA Energy Minister on Energy Transition at FII8," YouTube video, posted by "Future Investment Initiative," October 29, 2024

24. Matthew Martin, "Chip Startup Groq Backs Saudi AI Ambitions With Aramco Deal," Bloomberg, September 16, 2024, https://www.bloomberg.com/news/articles/2024-09-16/chip-startup-groq-backs-saudi-ai-ambitions-with-aramco-deal

25. Nino Orto, "Desalination: Miracle or False Hope for Water-Scarce Gulf Countries?" Fanack Water, May 28, 2024, https://water.fanack.com/publications/desalination-miracle-or-false-hope-for-water-scarce-gulf-countries/

26. Bindu Rai, "GCC Sovereign Wealth Funds Deployed $55B in First Nine Months of 2024," Zawya, October 1, 2024, https://www.zawya.com/en/wealth/funds/gcc-sovereign-wealth-funds-deployed-55bln-in-first-nine-months-of-2024-owa0qq1o

27. Middle East Economic Digest, "Breaking News: GCC Projects Market Breaks All-Time Spending Record," LinkedIn Post, November 22, 2023, https://www.linkedin.com/pulse/breaking-news-gcc-projects-market-breaks-all-time-spending-record-clbhf/

Rawan Oueidat and Ilya Tafintsev, "GCC Oil and Gas Producers' Capex Cooldown Won’t Fire Up Drillers’ Leverage," S&P Global, April 15, 2024, https://www.spglobal.com/ratings/en/research/articles/240415-gcc-oil-and-gas-producers-capex-cooldown-won-t-fire-up-drillers-leverage-13059990

Statista Research Department, "Military Spending in MENA by Selected Country 2023," Statista, July 19, 2024, https://www.statista.com/statistics/676109/mena-military-spending-by-selected-country/

Matt Smith, "Middle East Military Spend Shoots Up After Saudi Splurge," AGBI, April 27, 2023, https://www.agbi.com/analysis/economy/2023/04/middle-east-military-spend-shoots-up-after-saudi-splurge/

28. Andrew Homan, "Redefining Semiconductor Progress," Invest Like the Best podcast, November 5, 2024, https://joincolossus.com/episode/redefining-semiconductor-progress/

29. Ben Lefebvre, "Trump Calls on Russia and Saudi Arabia to Cut Oil Production," Politico, April 2, 2020, https://www.politico.com/news/2020/04/02/trump-calls-on-russia-and-saudi-arabia-to-cut-oil-production-161368

30. Julian Horsey, "Moore’s Law Squared: NVIDIA’s Jensen Huang’s Bold Vision for AI and Robotics," Geeky Gadgets, September 23, 2024, https://www.geeky-gadgets.com/moores-law-squared/

31. Sam Hillier, "AI Boom Drives Data Center Investments Amid Rising Risks," Transacted, September 13, 2024, https://transacted.io/ai-boom-drives-data-center-investments-amid-rising-risks/

32. Goldman Sachs, “Bullish Expectations for US Electricity Are Attracting New Power Traders," Goldman Sachs, July 18, 2024, https://www.goldmansachs.com/insights/articles/bullish-electricity-attracting-new-power-traders

33. Carly Davenport, Brian Singer, Neil Mehta, Brian Lee, and John MacKay, "Generational Growth: AI Data Centers and the Coming U.S. Power Surge”, Goldman Sachs Research, April 28, 2024, https://www.goldmansachs.com/pdfs/insights/pages/generational-growth-ai-data-centers-and-the-coming-us-power-surge/report.pdf

34. Jason O’Conaill, "Data Centres: A Cornerstone of Ireland’s Foreign Direct Investment and Economic Growth," Data Centres Ireland, accessed December 15, 2024, https://www.datacentres-ireland.com/data-centres-a-cornerstone-of-irelands-foreign-direct-investment-and-economic-growth-a-view-from-jason-oconaill-industry-expert/

35. Sacra Research, CoreWeave: Scaling AI Infrastructure for the Future, February 16, 2024, https://assets.ctfassets.net/f1df9zr7wr1a/5C0qxmXRFuOrWdCzKBehm6/ec6c4a22d10808e371139de572273530/coreweave.pdf

36. Christine Vuu, "Taiwan’s Semiconductor Strategy: A Developmental State Model for Economic Growth," CISLA Senior Integrative Projects, no. 72 (2024): 1-45, https://digitalcommons.conncoll.edu/sip/72

37. International Renewable Energy Agency (IRENA), "New Report Highlights Renewables as Climate Mitigation and Economic Diversification Opportunity for Gulf Countries," IRENA, December 6, 2023, https://www.irena.org/News/pressreleases/2023/Dec/Renewables-as-Climate-Mitigation-and-Economic-Diversification-Opportunity-for-GCC

38. H.R.H Prince Abdulaziz bin Salman Al Saud, "KSA Energy Minister on Energy Transition at FII8," YouTube video, posted by "Future Investment Initiative," October 29, 2024

39. World Nuclear Association, "Nuclear Power in the United Arab Emirates," World Nuclear Association, updated September 5, 2024, https://world-nuclear.org/information-library/country-profiles/countries-t-z/united-arab-emirates#:~:text=The%20UAE%20expects%20that%20the,via%20the%20regional%20power%20grid

40. Kunio Saito, "The Japanese Economy in Transition: Japan’s Adjustment to the Recent Worldwide Inflation, Recession, and Other Disturbances, and Its Plan for the Latter 1970s," IMF eLibrary, June 1, 1977, https://www.elibrary.imf.org/view/journals/022/0014/002/article-A010-en.xml