Nvidia at a crossroads after Huang’s exit

A fractured circuit board suspended in mid-air, one half glowing with intense golden light, the other dim and cracking, with thin wires stretching toward competing, shadowy structures in the distance.

Nvidia’s era of unchallenged AI chip dominance may be entering a new phase. Jensen Huang, the visionary co-founder who steered the company into the heart of the AI revolution, has stepped down following revelations of a personal relationship with an executive linked to Chinese intelligence. With CFO Colette Kress in temporary command, the company faces mounting technical, geopolitical, and structural challenges — just as rivals close in and the very tools Nvidia enabled begin to erode its advantages.

Leadership vacuum at a critical moment

Huang’s exit follows a decade of aggressive foresight. He funded early AI research, distributed free hardware to startups, and built CUDA — the software ecosystem that became a de facto standard for AI development. His departure, triggered by an affair with Alice Huang of Megaspeed — a firm reportedly importing $4.6–7.5 billion in Nvidia chips into Southeast Asia — has raised national security concerns in the US. While Nvidia remains a $215.94 billion giant, the timing of his fall is precarious. As one anonymous AI CEO told WIRED, chip designs in market today were conceived two to three years ago; the next wave depends on decisions made now. Leadership under a finance executive, rather than a technologist, risks shifting focus from innovation to margins — a trajectory analysts compare to Intel’s decline under CFO Bob Swan.

Competition closes in across markets

Nvidia still holds 80% of the AI accelerator market, but its grip is loosening. AMD’s share has grown from 2% in 2022 to 7–10% in 2026, bolstered by its acquisition of Taalas, a firm specialising in burned-in model weights for efficiency. Meanwhile, Google, Amazon, and Microsoft are designing custom silicon to cut costs and reduce dependency. Startups like Cerebras offer alternative architectures for large-scale AI, while Wafer, an AI-driven porting service, claims it can adapt major open-weight models to run on AMD hardware at 80% efficiency within hours. This undermines Nvidia’s CUDA moat: if AI can automatically rewrite code for rival chips, software lock-in weakens.

China’s rise and Nvidia’s shrinking footprint

In China, Nvidia’s influence has sharply declined. US export controls bar sales of its most advanced chips, and Beijing is accelerating domestic alternatives. Huawei has developed supercomputers capable of training top-tier AI models, and firms like Z.ai report full reliance on Chinese silicon. While Nvidia’s sales of older chips in China doubled year-on-year, the region now contributes just 8% of revenue, down from 20% in 2024. Geopolitical tensions have also tarnished Huang’s pragmatic stance on China engagement — a stance once seen as commercially sound but now viewed with suspicion. Analysts at Eurasia Group note that cutting off China risks strengthening local competitors, a prediction now playing out.

AI begins to disrupt chip design itself

Ironically, the AI tools powered by Nvidia’s hardware are now automating chip design. Firms use AI to optimise processor layouts, and models from OpenAI and Anthropic can generate electronic designs from scratch. While Nvidia has access to these tools too, the broader trend threatens its ecosystem advantage. AI’s ability to rapidly port models across hardware platforms reduces switching costs. Wafer’s CEO Emilio Andere says his engineers now do little more than monitor AI-generated code. This accelerates commoditisation — a danger for any dominant platform.

What it means

For businesses relying on Nvidia, the shift demands diversification. Cloud providers investing in custom silicon signal a long-term move away from vendor lock-in. Startups should explore AMD and alternative architectures, especially as AI lowers the barrier to porting. In the UK, UAE, US, and Kazakhstan, firms should assess supply chain resilience, particularly around US-China tech restrictions. OpenAI CEO Sam Altman says the industry will be “fine” without Huang — but that confidence rests on sustained demand, not structural stability. Nvidia remains powerful, but its moats are narrowing.

The company’s next leader will shape not just its future, but the balance of power in AI infrastructure. If innovation slows, competitors are ready to fill the gap.

Sources

Leave a Reply

Your email address will not be published. Required fields are marked *