The Trump-Xi Beijing summit produced its first concrete AI outcome this week. In a Reuters exclusive picked up by CNBC and multiple outlets on May 14, the US government confirmed it has cleared roughly 10 Chinese technology companies to purchase Nvidia’s H200 AI accelerator chips — the second-most powerful chip in Nvidia’s current lineup.

The approved firms include Alibaba, Tencent, ByteDance, and JD.com. Lenovo and Foxconn have been authorized as distributors. Each approved buyer is allowed to purchase up to 75,000 chips, either directly from Nvidia or through those designated intermediaries.

The catch: not a single chip has shipped.

The Stall

Despite the US government approvals, deals remain frozen. According to reporting from Reuters, Beijing has quietly signaled to major tech platforms that they should slow or redirect their Nvidia orders toward domestic chipmakers — in particular, Huawei’s Ascend line of AI accelerators.

This is a rare situation where both governments have technically said yes, and the transaction still isn’t happening. US firms now have approval to sell. Chinese firms have approval to buy. But Beijing’s informal pressure is operating as an effective block on the other side.

Nvidia CEO Jensen Huang traveled to Beijing with Trump’s delegation — reportedly joining the trip mid-flight, picked up in Alaska en route. His presence in the room signals how central the chip question is to both companies and governments navigating this moment.

The Broader Trade Context

The H200 clearance follows the US-China trade truce announced on May 12, which cut US tariffs on Chinese goods from 145% to 30% and Chinese tariffs on American goods from 125% to 10%. AI chips are now part of what both governments are treating as a managed trade relationship rather than a full-scale technology war.

That matters for how businesses should think about AI supply chains. The previous period — when chip restrictions were tightening almost quarterly — was marked by unpredictability. The current direction suggests both governments are trying to establish a floor under the relationship, not necessarily a ceiling on competition.

Under the terms of any approved deals, the chips must physically transit US soil before being re-exported to China. The US government also collects 25% of the revenue from these chip transactions.

What This Means for Business

AI compute access is a geopolitical variable. The H200 clearance is a reminder that access to the hardware that powers AI is not purely a commercial question. It is shaped by US export control policy, domestic Chinese industrial strategy, and the state of US-China relations at any given moment. Businesses building AI systems on infrastructure that depends on this supply chain need to factor that instability into their planning.

Chinese AI development is not being stopped. Even with the stalled deliveries, the approval of 75,000 chips per firm represents significant compute capacity. If those deals eventually close, Chinese AI development will accelerate. For businesses competing in markets where Chinese AI tools are or will be relevant, the competitive landscape is changing.

Huawei is a bigger variable than it gets credit for. Beijing steering platforms toward Huawei Ascend chips rather than Nvidia H200s is not just a trade posture. If Huawei’s hardware closes the performance gap, it changes the global AI chip market dynamics. It also creates a world where two AI hardware ecosystems — one anchored by Nvidia, one by Huawei — run in parallel across different geographies. That fragmentation has real implications for companies operating internationally.

For Australian and Asia-Pacific businesses, this matters. Enterprise DNA works with companies across APAC. The chip clearance and the trade dynamics around it will directly shape the availability and pricing of AI infrastructure across the region over the next 12 to 24 months. Businesses planning significant AI investment need to track this, not just the model releases.

The Larger Shift

The presence of Jensen Huang, Tim Cook, Elon Musk, and other top US tech executives in Beijing for the Trump-Xi summit is itself a data point. When the CEOs of the world’s largest AI and technology companies travel to diplomatic summits, it reflects how central AI infrastructure has become to the relationship between nations.

The prior narrative was that AI was fracturing into two incompatible ecosystems — one American, one Chinese. What the summit suggests, at least tentatively, is that both governments are looking for some level of managed engagement rather than complete bifurcation. That does not mean alignment. But it means the situation is more fluid than the last 18 months of tightening restrictions implied.

For businesses making AI investment decisions, fluidity is better than hostility — but it still requires active monitoring rather than passive assumption that supply chains will stay stable.

---

Enterprise DNA helps businesses navigate AI adoption — from strategy to implementation. If your organisation is evaluating AI infrastructure or workforce decisions in a shifting landscape, book a discovery call with our team.