Report: Who Is Really Winning the Global AI Race, With 725 Billion Dollars of Spending Planned for 2026 and the Model Gap at 2.7 Percent
The Edge for Economic Consultancy has published a special report, “The 725 Billion Dollar Question: Who Is Really Winning the Global AI Race?”, examining who leads across the five layers of the artificial-intelligence economy. The report finds no single winner but a contest increasingly defined by capital and infrastructure: Amazon, Microsoft, Alphabet and Meta alone plan between 695 and 725 billion dollars of capital spending in 2026, a 77 percent increase on about 410 billion dollars in 2025, even as the gap between the best US and Chinese models has narrowed to about 2.7 percent by March 2026.
The report’s central argument is that the AI race is not one contest but five interlocking layers, frontier models, consumer distribution, enterprise adoption, cloud infrastructure and semiconductors, each with a different leader, and that victory now turns on capital scale, infrastructure and physical constraints rather than model benchmarks alone.
On the frontier, the report finds the best US and Chinese models have converged to a gap of about 2.7 percent by March 2026, from a range of roughly 17.5 to 31.6 percentage points in May 2023, even as US private investment, at 285.9 billion dollars in 2025, still dwarfs China’s 12.4 billion dollars. The narrowing capability gap alongside a widening spending gap is, in the report’s reading, the defining tension of the race.
Leadership splits by layer. The report places OpenAI ahead in consumer reach with more than 900 million weekly users, Anthropic scaling fastest in enterprise at a 47 billion dollar revenue run-rate, Google holding the most integrated stack from chips to models to cloud, and Nvidia as the largest financial beneficiary with 75.2 billion dollars of data-centre revenue in a single quarter. No company leads every layer.
For the Gulf, the report identifies a widening role. It highlights the one-gigawatt Stargate cluster within a broader five-gigawatt campus in Abu Dhabi, with a first 200-megawatt phase scheduled for 2026, and notes an artificial-intelligence adoption rate of 64 percent of the population in the United Arab Emirates against 28.3 percent in the United States. It points to Saudi Arabia’s HUMAIN initiative and its emphasis on Arabic-language systems, and sets out opportunities in financial services, government platforms, energy optimisation and data hosting for Kuwait and other GCC economies.
Why it matters: The report reframes the AI race from a contest of model benchmarks to one of capital, infrastructure and monetisation, and its central question, whether spending approaching 725 billion dollars a year will earn its return, bears directly on the technology exposure in Gulf sovereign and institutional portfolios and on the region’s own data-centre ambitions. Its conclusion, that a layered ecosystem is more likely than winner-takes-all concentration, favours firms that combine computing capacity, distribution, customer relationships and enough cash flow to renew infrastructure continuously.
| Indicator | Latest | Context |
|---|---|---|
| Planned 2026 capital spending, four hyperscalers | 695 to 725 billion dollars | Up 77 percent on about 410 billion in 2025 |
| Best US-China model gap, March 2026 | About 2.7 percent | From 17.5 to 31.6 points in May 2023 |
| US private AI investment, 2025 | 285.9 billion dollars | Versus 12.4 billion dollars in China |
| OpenAI weekly users | More than 900 million | Consumer-distribution leader |
| Anthropic revenue run-rate | 47 billion dollars | Fastest enterprise scaler |
| Nvidia data-centre revenue | 75.2 billion dollars | Single quarter, largest beneficiary |
| UAE AI adoption | 64 percent of population | Versus 28.3 percent in the United States |
Download the English report | Download the Arabic report
Outlook: The report identifies the tests ahead as utilisation, pricing discipline and whether AI produces measurable new productivity rather than merely shifting existing software spending. As capabilities converge, it argues, differentiation will shift further from training the largest model toward the cost and reliability of delivering useful intelligence, the terrain on which the next phase of the race will be decided.
Sources: The Edge for Economic Consultancy; company disclosures.
