About a year ago, I wrote a post calling the Middle East – specifically Saudi Arabia, Qatar, and the United Arab Emirates – the “swing vote” of the global AI competition between the US and China. Last week, it looks like this swing vote has swung decidedly towards the US, at least for now.
Trump’s four-day swing through the three countries – the first major foreign trip of his 2nd term – had all the relevant CEOs of leading AI companies in tow, ready to take purchase orders. The Middle Eastern kings and royal families, having felt sufficiently wooed by the salesman-in-chief, committed hundreds of billions of capital and gigawatts of energy to buy American-designed, though most likely Taiwan-made, technology products.
Cheap capital and energy are two of the Middle East's most powerful advantages. This leverage has been flexed in spades between the US and China. The fact that the US “won” this round is all too predictable. When I first analyzed this dynamic last year, I called China’s efforts in the region “more aggressive and tangible”, while America’s (under a different president at the time) was “lukewarm”. But given that American companies have the superior products in AI, all it would take is enough lobbying from these companies of a different president, who loves big numbers, big deals, and willing the drop moral judgments on the Arab world in exchange for those deals – in other words, more like China – for the region to swing.
That’s exactly what happened last week. (Meanwhile, Israel’s own announcement of a national AI supercomputer to support its much more advanced tech ecosystem was overshadowed and received scant coverage. The company that will build this supercomputer, Nebius, is a company we have done a lot of research on if you want to learn more.)
There has been, and will continue to be, lots of controversies and debates on the details of these deals as they get fleshed out into reality. But this US-Middle East AI alliance reveals two important long-term truths for me about how AI will spread globally in the next decade.
AI Is Just Another Technology
Ironically, the massive AI infrastructure deals struck between the US and the Middle East shows that AI is just another general purpose technology and is not that scary. It may be special, but it is not that special. It is certainly not existentially special enough to warrant being compared to nuclear weapons or the Manhattan project. Otherwise, how can it be sold willy-nilly to any country, friend or foe, when that kind of world-destructive power should be kept for yourself!
The Internet made global interconnectedness possible. Mobile turbo-charged social media. Cloud computing catalyzed the increasing digitization of everything, including generative AI (ChatGPT would not be possible without the cloud). So AI will change a lot of things, just like its technology platform predecessors, some for the better, some for the worse, but the world won’t end because of it.
Couching AI with a more moderate risk profile also shows the path of US-China competition as one that will be focused mostly on economic competitiveness, with some military implications, but nothing world-ending. Whether the US or China (or now the Middle East) reaches the poorly-defined end state of AGI first is really not that consequential and by no means a human extinction level event. Because if you invert the situation, and AGI is an end state of extinction level significance, then why would you ever sell the secret to that future (aka billions worth of NVIDIA GPUs) to any other country? After all, even the Trump administration, which wants and respects hard power above everything else, wouldn’t be so careless and naive as to let this ultimate power rest somewhere else…right?
The US-Middle East AI alliance definitively shows that "democratic AI” or “American AI”, often used interchangeably these days, is nothing more than a highly effective sales pitch for GPU makers and cloud hyperscalers. As these GPUs come online inside the domains of Middle Eastern monarchies, the probability of democratic AI “enlightening” the Arab royal families to be a bit less…monarchic is basically zero.
In fact, the net positive effect of AI spreading democratic values will pale in comparison to the early days of the Internet, where information, culture, and different ways of life more freely spread around the world than at any other time in the past. Why? Because the nature of AI infrastructure and diffusion is centralized, controlled, not organically bottoms-up, and sometimes even requires presidential intervention to make happen. And sooner or later, a general purpose technology bends to the will of its buyer and never the other way around, whether it’s Cisco routers or NVIDIA GPUs.
Swing Votes Have No Loyalty
The other long-term implication, or more of a truth of human behavior, is that swing vote never stops swinging.
Team America may have resoundingly won the Middle Eastern AI vote now, but for the Middle East to maintain its leverage in the global AI competition, it cannot be tied to the American hip. A swing vote’s inherent power comes from no loyalty; disloyalty is part of the swing vote formula. Expect G42, HUMAIN, Core42, Datavolt, and other swing vote beneficiaries to be high-maintenance customers, as they should be if only to stay relevant.
A successful swing vote also spawns imitators. What the Middle East has accomplished with Trump offers important tips and ideas for other regions that crave the same leverage and power in AI. The Middle East’s three important advantages I previously laid out – money, energy, geography – aren’t unique. There are other petrostates with large government coffers, a central location on the globe, and enough of an authoritarian instinct to quickly permit energy sources and break grounds on large projects to attract the attention of "democratic AI” vendors to fill their next quota.
After all, reaching the mystical status of a swing vote and being politically relevant is surprisingly arbitrary. Take New Hampshire, for example, a tiny and hardly representative state in the upper Northeastern corner of the US, where I have lived and done campaign work before. Followers of US domestic politics would know that it has been a perennial swing vote in presidential elections for many cycles, though its influence during the general election has waned a bit in recent years. However, it has always maintained a powerful presence in the primary stage of a US presidential election – the process where each political party selects its candidate – as the first primary state to vote on the calendar. How did New Hampshire get to be this powerful and punch way above its weight? Because the New Hampshire constitution stipulates that its primary election must be held “at least seven days earlier than any similar election held in any other state” for the sole purpose of protecting the “tradition of the New Hampshire first-in-the-nation presidential primary.” In other words, New Hampshire is a swing vote and a politically powerful state in America simply because its own state constitution said so!
So if you are Brazil or Malaysia or even Nigeria, you too can be a swing vote simply because you said so, and in the context of AI, be willing to back it up with capital, energy, and a healthy dose of disloyalty.