NVIDIA and TSMC - first Blackwell chip
Nvidia CEO Jensen Huang (centre) celebrates the first US-made Blackwell AI chip wafer in Arizona, a key step in securing America's tech lead. Nvidia

Nvidia has reported a staggering $57 billion (£43.69 billion) in quarterly revenue, driven by its new US-made AI chips. This record-breaking figure underscores the widening technology gap between the US and China, amid ongoing sanctions that leave Beijing years behind in advanced chip development.

These results do more than boost investor confidence; they serve as clear evidence of the growing chasm between the two superpowers. The US's strategy of heavily investing in domestic AI infrastructure while restricting China's access to top-tier technology is now bearing fruit. It paints a stark picture: one side is building the future, while the other is forced to play catch-up.

The US 'AI Factory' Boom

At the heart of Nvidia's success is its Data Centre division, which alone generated a record $51.2 billion (~£39.2 billion). This surge is driven by the company's latest Blackwell chips, which are now central to America's domestic manufacturing push.

A significant milestone was achieved with the first Blackwell wafer produced at TSMC's new Arizona facility. Nvidia's founder and CEO Jensen Huang described this as creating a 'virtuous cycle of AI', where rising demand fuels investment in more powerful technology, accelerating AI development.

This cycle is driving a wave of major infrastructure projects across the US. The Department of Energy is collaborating on Solstice, an AI supercomputer set to utilise 100,000 Blackwell GPUs. Meanwhile, Elon Musk's xAI is constructing a colossal data centre in Memphis designed to house over half a million chips.

Why China Is Falling Behind

China, however, is noticeably absent from this AI gold rush. The reason is clear: US export controls. Washington has effectively blocked Chinese firms from purchasing Nvidia's advanced chips, forcing reliance on domestic alternatives.

Huawei's Ascend 910B processor, for example, is the primary alternative. Yet, a July report from think tank MERICS suggests its performance is roughly comparable to Nvidia's A100, a chip released in 2019.

The hardware gap is only part of the challenge. The bigger obstacle is software. Nvidia's CUDA platform has long served as the industry standard, facilitating development and interoperability across its chips. Huawei's CANN software, in contrast, is like asking developers to learn a new, less reliable language from scratch. The report notes stability issues that deter companies from switching, with Nvidia's restricted H20 chip still outselling Huawei's top offering in China by more than five-to-one in 2024.

This dynamic illustrates the current bifurcation of the AI landscape. The global industry is now effectively split into two competing technology 'stacks', each built on limited, contested foundations of chipmakers, as depicted in MERICS' diagrams.

MERICS Infographics: AI Stack
Diagram showing the AI 'stack' highlights the global chip duopoly: Nvidia and TSMC versus Huawei and SMIC in China. MERICS

The Accelerating AI Arms Race

Nvidia's record quarter is a clear victory for Washington's dual strategy: investing heavily in its own AI sector while actively restricting China's access to advanced technology. This has created two distinct ecosystems.

On one side, a US-led network of private corporations, cloud providers, and government labs collaborate on a shared, market-leading platform. On the other, China's state-driven approach, led by Huawei, aims to rebuild an entire tech stack from scratch under intense pressure.

The disparity is now measurable in real-time through quarterly earnings, supercomputer deployments, and the widening technology gap. Nvidia's results signal that the global AI arms race has entered a new, accelerated phase—one driven by strategic geopolitical and economic factors that will shape the future of technology development.

Writer's pick