Nvidia CEO: Taiwan Semiconductor Pivotal for AI Chips

Taiwan's Enduring Centrality in the AI Supply Chain

In a compelling statement that cuts through ongoing debates concerning global semiconductor production, Jensen Huang, CEO of Nvidia, has unequivocally declared that the trajectory of advanced chipmaking continues to run through Taiwan. This assertion firmly positions Taiwan Semiconductor Manufacturing Company (TSMC), the undisputed leader in the chip space, as the enduring cornerstone of the industry, irrespective of the numerous fabrication plants announced across other regions.

Huang's insights, recently shared in an interview with The Times (UK), underscore a crucial distinction: efforts towards diversifying chip production should be perceived through the lens of resilience and strategic insurance rather than as a viable substitute for Taiwan's deeply entrenched capabilities. While the past year has seen significant capacity expansions in the United States and Europe, the complex, multi-decade ecosystem essential for producing cutting-edge silicon remains predominantly concentrated within Taiwan.

This perspective is particularly salient given the substantial investments made globally. The Semiconductor Industry Association’s report from July last year highlighted over 100 projects in 28 U.S. states, attracting more than half a trillion dollars in private investment, with U.S. chip capacity projected to triple by 2032. However, as Huang articulates, these initiatives primarily serve to bolster supply chain security rather than to replicate Taiwan's unique advantages.

The Evolving AI Market Landscape and Investor Sentiment

Amidst these strategic industry shifts, the Artificial Intelligence (AI) trade continues its dynamic evolution, prompting investors to adopt a more selective approach. Recent market movements illustrate this trend vividly. Memory stocks, for instance, have shown remarkable strength, with Micron Technologies experiencing nearly a 10% gain, buoyed by the burgeoning demand-supply imbalance within its sector.

TSMC stock, while maintaining a steadier short-term trajectory, has delivered an impressive almost 40% gain over the past six months, reflecting its critical role in the AI ecosystem. Zooming out, the broader semiconductor space, as represented by the SOXX ETF, recorded a robust 5% gain over the past week. This surge indicates strong investor confidence in "picks-and-shovels" companies, such as Applied Materials, which provide essential equipment and services to chip manufacturers.

Conversely, established AI bellwethers like Nvidia, despite consistently showcasing ambitious innovations at events such as CES, have recently exhibited a somewhat sluggish performance in the immediate term. Following its coverage on January 7, 2026, Nvidia's stock experienced a modest decline from $189.11 to $184.86 by January 9, 2026, marking a 2.25% drop. Nevertheless, the underlying market dynamics suggest that the AI trade remains robust, with Huang’s recent commentary further energizing market sentiment.

Debunking the AI "Bubble" Narrative: A Foundational Shift

Concerns regarding an "AI bubble" are growing louder, fueled by surging spending figures and skyrocketing stock valuations. Nvidia, for example, boasts a staggering market capitalization of approximately $4.5 trillion, having added over $1 trillion since the beginning of 2025 alone. Despite these eye-popping figures, Huang contends that such concerns misinterpret the current landscape, suggesting that what appears to be excess is, in fact, the nascent stage of a profound paradigm shift in computing itself.

Indeed, the scale of AI spending is monumental, evidenced by the latest capital expenditure (capex) figures from the "Big 3" hyperscalers:

• Amazon (AWS): CFO Brian Olsavsky projected full-year 2025 capex to reach nearly $125 billion, with expectations for even higher spending in 2026, largely attributed to AI-driven projects.
• Microsoft (Azure): Microsoft reported nearly $35 billion in capex during its fiscal Q1 2026, signaling continued increases as the company expands its AI data-center capacity.
• Alphabet (Google Cloud): Alphabet significantly revised its 2025 capex guidance upwards to a remarkable $91 billion to $93 billion, directly linked to surging AI and cloud demand.

Huang further clarifies this shift by describing AI as a wholesale transition from general-purpose computing to accelerated computing. This transformation is spearheaded by sophisticated AI agents, models, and applications that operate atop this advanced foundation, fundamentally altering how computational tasks are executed and optimized.

TSMC's Unrivaled Position and Pricing Power

Beyond the sheer volume of investment, Huang also addresses concerns that Nvidia’s extensive ecosystem investments might introduce systemic risks. In the current environment, TSMC effectively controls critical bottlenecks within the AI supply chain. This unique and powerful positioning grants TSMC substantial pricing power, which is increasingly reflected in its robust fundamentals.

As of Q3 2025, High Bandwidth Memory (HBM) represented a massive 57% of TSMC's sales, with its cutting-edge 3nm nodes accounting for 23% of wafer sales, merely a few quarters into their ramp-up phase. Furthermore, with the impending ramp-up of 2nm wafer production, expected to be priced over $30,000 per wafer, combined with 5%–10% price hikes on sub-5nm nodes, major customers such as Nvidia and Apple face limited alternatives but to accept these terms. This scenario solidifies TSMC’s robust strategic positioning as the indispensable backbone of the global AI buildout, rendering it less susceptible to short-term market cycles or "bubble" debates.