Part IV: Scaling, Capital, and Rules
Chapter 22
AI Around the World
On 12 May 2025, the eve of President Trump's visit to Riyadh, Saudi Arabia's Public Investment Fund launched HUMAIN, a wholly owned national AI company chaired by Crown Prince Mohammed bin Salman and run by Tareq Amin, the former chief executive of Aramco Digital. Its mandate was the whole stack: data centres, cloud, Arabic language models, applications and investment (company). The next day, at the US–Saudi Investment Forum, it signed with almost every major American supplier: Nvidia, for 18,000 GB300 chips to start and several hundred thousand over five years; AMD, for about $10 billion and 500 MW over five years; AWS, for an "AI Zone" worth more than $5 billion; and Qualcomm, for a memorandum on data-centre chips (companies). On the same day, 13 May, Washington announced that it would rescind the diffusion rule that would have capped advanced chips for countries like Saudi Arabia, as Chapter 21 describes.
Six months later, on 19 November 2025, HUMAIN announced that Nvidia would supply "up to 600,000" GPUs over three years and AWS up to 150,000 in Riyadh, and around 19–20 November the Commerce Department approved HUMAIN and the UAE's G42 each to buy up to about 35,000 Blackwell-class chips (companies, reported). The gap between the two numbers was the story. At about 1.4 kW per GPU including overhead, 35,000 chips draw about 50 MW and 600,000 about 840 MW (stylized arithmetic). By late 2026 Saudi Arabia was reported to be reining in spending, and in September HUMAIN was seeking about $2.5 billion of outside capital for a data-centre fund and assembling a team to prepare for a listing (Bloomberg, Fortune, reported).
This chapter looks at AI outside the US and China: what different countries have tried to own, and what it has cost them.
The preview ends here. The full chapter is in the book.
In this chapter
- 22.1The Gulf: chips for alignment
- 22.2India: a domestic layer for a price-sensitive market
- 22.3Europe: one champion and a consolidation
- 22.4Asia and Africa: specialists and exits
- 22.5A worked example: owning a national cluster
- 22.6How an expert reads a national AI strategy