AI: Nvidia’s Golden Ticket

Nvidia’s meteoric ascent is tightly linked to its dominance in advanced microchips designed for the accelerating demands of AI. Its cutting-edge graphics processing units (GPUs) are essential hardware for training and running large language models—the backbone of AI services used by companies such as OpenAI, Google, and Amazon. These chips underpin data centers worldwide, enabling applications from real-time translation to autonomous vehicles and generative AI assistants like ChatGPT.

In its latest earnings report (Q2 2025), Nvidia announced revenues of $46.7 billion (£34.5 billion) for the past three months, marking 56% year-on-year growth—astonishing momentum for a business of this scale. The company’s net profit for the quarter reached an unprecedented $26.4 billion (£19.5 billion), with gross margins topping 70%. With results like these, Nvidia’s shares have soared, leaving traditional tech peers such as Apple, Alphabet (Google), and Meta (Facebook) trailing in market capitalisation.

Jensen Huang, Nvidia’s founder and CEO, reinforced the bullish mood calling this the “boom time for AI.” The company projects even stronger results for the coming quarter, forecasting sales of $54 billion (£40 billion), even excluding any potential revenue from China due to ongoing export restrictions.

AI Boom or Bubble?

Yet even as Nvidia’s financials dazzle, questions swirl about the sustainability of the AI gold rush. Skeptics see worrying parallels to the late 1990s, when explosive investment in internet infrastructure left many companies over-leveraged and led to spectacular collapses by 2001.

“Are we in a phase where investors are overexcited about AI? My opinion is yes,” Sam Altman, CEO of OpenAI, said last month. “Is AI the most important thing to happen in a very long time? My opinion is also yes.”

Industry analysts are concerned about the vast sums being poured into AI data centers and infrastructure. Harris Kupperman, CEO of Praetorian Capital Management, estimates that global spend on AI data centers will top $400 billion in 2025. He notes that to simply recoup 2025’s investment, AI companies would need to generate $160 billion in annual revenue for the next decade—a tenfold jump from current levels.

The fear is that in the race to not miss out on the AI revolution, tech companies and investors are dramatically overbuilding capacity and ordering far more Nvidia chips than necessary. Such overcapacity could lead to an inevitable shakeout if AI growth fails to meet today’s lofty projections.

Historical Parallels: The Dotcom Warning

The situation echoes the telecoms and internet boom of the late 1990s. During that era, companies spent vast sums on laying fibre-optic cabling and building internet infrastructure—outlays that proved essential over the long term, but decimated the short- and medium-term valuations of the firms involved. Global Crossing, at one time a darling of Wall Street for its fibre-optic network investments, lost 97% of its peak valuation within just two years.

Kupperman draws direct parallels between today’s AI datacenter spending—driven by Nvidia’s hardware—and the dotcom excess. “I really thought the CEOs of today, educated by the prior cycle, would never repeat the mistake of overbuilding at massive scale without revenue. Yet, here we are again,” he commented in a recent analysis.

Shifting Geopolitical Landscape and Supply Chain Risks

Nvidia’s supply chain and geopolitical position add further complexity. The United States, concerned about China’s rapid progress in AI and supercomputing, has repeatedly moved to restrict the export of Nvidia’s most advanced chips to Chinese firms. With China representing a significant portion of demand for AI hardware, these rules create uncertainty for Nvidia’s growth trajectory.

The company’s critical chip manufacturing partners are based in Taiwan, which produces the world’s most advanced semiconductors at foundries like TSMC. Taiwan’s strategic importance—not to mention its fraught political relationship with mainland China—places Nvidia at the nexus of global technology and trade tensions. Notably, CEO Jensen Huang, who is Taiwanese-American, has been involved in negotiations about possible deals to allow chip exports to China, including controversial proposals that might see the US government taking a share of related revenues.

Although Nvidia projects robust revenue even if China is excluded from the equation, the company’s fortunes remain vulnerable to political shocks, trade tit-for-tat, and supply chain disruptions.

What’s Next for Nvidia and AI?

For now, optimism prevails—AI is being rapidly integrated into enterprise software, consumer services, financial modelling, and even creative industries. Major cloud providers and tech giants continue to order Nvidia’s cutting-edge H200 and Blackwell GPUs by the container-load, a sign “AI winter” is not yet in sight.

Nvidia’s ability to maintain its technical lead—amid growing competition from rivals like AMD, Intel, and upstart AI chip firms—will be crucial. Meanwhile, some of the AI startups buying Nvidia hardware by the pallet may find themselves squeezed if revenue fails to materialize as forecasted, potentially spelling trouble for the broader AI ecosystem, and, by extension, Nvidia’s future sales growth.

Yet, even if the current pace of growth slows, Nvidia’s hardware will remain an indispensable foundation for AI development, just as fibre-optic networks undergirded the internet despite the 2001 crash. The company, now at the heart of both the technology and trade battles of the 21st century, faces an era of extraordinary opportunity and no shortage of risk.