AI chip export curbs risk weakening US competitiveness, Pilkington argues

Restricting advanced AI chip exports to China could weaken US companies’ competitiveness, Philip Pilkington argued, saying hardware competition and industrial productivity gains matter more than preserving a technological gap through export controls.

In a discussion on Mario Nawfal’s X account on November 10, Pilkington backed Nvidia CEO Jensen Huang’s case for allowing chip exports to China and described an approach centered on restrictions as economically misguided.

Pilkington identified hardware, including Nvidia’s GPUs, as an area of US strength. He explained Huang’s position as maintaining access to China to preserve market share while continuing to develop leading chips. Supporters of restrictions, by contrast, argue that denying China advanced GPUs would keep it several years behind and help the US win the AI race, he said.

He linked that disagreement to competing AI software business models. Closed-source models pursued by OpenAI and Anthropic restrict access to their internal workings and charge for use, he said, while Chinese open-source models such as Deepseek give users greater scope to use and modify the technology.

Pilkington acknowledged that China’s reasons for favoring open-source models were uncertain. His interpretation was that the emphasis lay less on monetizing software itself and more on applying AI in factories to improve productivity. He considered that approach economically stronger.

To explain the limits of export controls, Pilkington compared the spread of AI and chips with the printing press. Attempts to suppress a technology cannot prevent others from developing alternatives, he argued, limiting the effectiveness of restrictions as a way to manage competition with China.

Pilkington also expressed concern about the Nasdaq’s vulnerability, arguing that purchases of Nvidia chips and their installation in data centers were central drivers of the market.

He questioned whether shifting emphasis toward subscriptions for closed-source AI software could replace that investment activity, underscoring his view that hardware spending remains fundamental to the AI market.