← Front Page
AI Daily
Geopolitics • Thursday, 24 September 2026

Xi Comes to Washington, and AI Is the Fault Line

By AI Daily Editorial • Thursday, 24 September 2026

State visits usually mark warm relations. This one marks a rivalry. Xi Jinping arrived in Washington this week for his first visit in a decade, sitting down with Donald Trump for a two-day summit that opens Thursday, and by every account the defining subject is not trade or Taiwan but artificial intelligence. "Whoever wins AI wins," Trump has said, more than once. Neither side expects a grand bargain. Both want to look steady while circling the thing they actually care about.

The competitive framing is unmistakable on both ends. The United States still leads on the most advanced models and chips, and it has tried to protect that lead by choking off exports of the most powerful processors, while accusing Chinese developers of siphoning capability from American models at industrial scale. Beijing calls the export controls an attempt to monopolize the field. In a blunt essay, China's minister of state security, Chen Yixin, warned that AI in hostile hands poses a direct threat to Communist Party rule, a rare admission that the leadership sees the technology as a matter of regime survival, not just economics.

Then, days before Xi landed, Alibaba changed the mood music. At its conference in Hangzhou the company unveiled the Zhenwu V900, which chief executive Eddie Wu called the most powerful AI chip in China today, claiming three times the performance of its predecessor and mass production in early 2027. Alibaba also said it plans to train a model in the range of five to ten trillion parameters and to build past twenty gigawatts of computing by 2032. Huawei showed new chip technology the week before. The timing was not subtle: China wanted to walk into the summit demonstrating that export controls have pushed it toward self-reliance rather than surrender.

The guest list underlined how much of this is really about industry. Among those attending the state dinner are Nvidia's Jensen Huang, OpenAI's Sam Altman, AMD's Lisa Su, Elon Musk and Tim Cook. Su, announcing her attendance, called the summit "very impactful" and argued that technology companies must keep pushing aggressively while building "trust in AI," which she called non-negotiable. These are the people whose products the two governments are fighting over, seated at the table where the fight plays out.

Amid the friction, there is one narrow patch of common ground. After talks in New York, Treasury Secretary Scott Bessent said the United States had proposed a "notification mechanism" so the two largest AI powers could warn each other about incidents that could affect national security. It builds on an earlier, limited agreement that decisions over nuclear weapons should stay with humans, not machines. Neither is a treaty. Both reflect a shared, if unspoken, worry that a powerful system slipping loose is a problem no export control can contain.

Beyond the summit, the two countries are already building rival camps. Washington launched Pax Silica late last year to secure technology supply chains among partners such as Japan and South Korea; Beijing countered with a cooperation body aimed largely at the Global South, pitching cheaper open models to developing nations. Southeast Asia is being courted by both. The irony, as one analyst put it, is that the two stacks can be mixed and matched: countries are already running Chinese open models on Nvidia chips. The real question the summit will not settle is whether either power can force the rest of the world to choose.

Sources