This Time is NOT Different.

Semiconductor Stocks, AI Capex, and Potential Economic Risks

Key Concepts:

• Asset Bubble: A market phenomenon characterized by a rapid and unsustainable increase in asset prices, exceeding intrinsic value. Defined in the video as requiring a >100% price increase over 2 years, >100% outperformance of the broader market over 2 years, and >50% 5-year return.
• AI Capex: Capital expenditure specifically directed towards Artificial Intelligence infrastructure (data centers, cloud computing, semiconductors).
• Multiplier Effect: The phenomenon where an initial injection of spending into the economy leads to a larger overall increase in economic activity.
• Labor Productivity: A measure of economic output per unit of labor input.
• Total Addressable Market (TAM): The total market demand for a product or service.

Historical Bubbles & Current Market Context

The National Bureau of Economic Research (NBER) defines an asset bubble using three criteria: a price increase exceeding 100% over two years, outperformance of the broader market by at least 100% over the same period, and a 5-year return exceeding 50%. Historical examples meeting these criteria include the railway mania of the 1840s, the 1920s Dow Jones boom, and the Japanese stock market in the 1980s – all of which were followed by significant economic downturns, including the Great Depression.

While mainstream media suggests a current tech bubble, the NASDAQ 100 does not meet the NBER’s definition, having risen 45% in the last two years, outperformed the S&P 500 by only 3%, and delivered a 90% return over five years. This contrasts sharply with the dot-com bubble, where the NASDAQ 100 met all three criteria. However, semiconductor stocks do meet all three criteria, exhibiting a 110% rise over two years, 275% over five years, and outperforming the S&P 500 by nearly 100% over the last two years.

The Significance of Semiconductors & AI Infrastructure

Although semiconductor stocks currently represent a smaller portion of the economy than tech did in 1999, their importance lies in their central role in the AI infrastructure buildout. Data centers, cloud computing, and large language models all fundamentally rely on semiconductor chips. A bubble in this sector could indicate a broader bubble encompassing the entire AI ecosystem.

AI capital spending (Capex) is already significantly impacting US growth. In 2025, it amounted to approximately $375 billion (1.2% of GDP) and is projected to more than double to $875 billion (3% of GDP) this year. This spending isn’t just a direct injection of capital; it generates a multiplier effect. For example, $177 billion in tech company AI Capex in the first half of 2025 reportedly generated $700 billion in US economic growth, representing a 3.5x multiplier.

Taiwan’s Role & Potential Risks

Taiwan’s chip export data is strongly correlated with US economic growth. Periods of accelerated Taiwanese chip exports coincide with strong US growth, while slowdowns in exports correlate with weaker US growth. Currently, Taiwan produces over 60% of the world’s semiconductors, making it a critical component of the global supply chain. A decline in Taiwanese chip exports could therefore negatively impact US economic growth.

Looking ahead to 2030, AI Capex is projected to reach $7 trillion, roughly 20% of current US GDP. Applying the observed 3.5x multiplier effect, this could potentially double the US economy by 2030. However, this projection is considered unrealistic, mirroring the formation of bubbles where expectations become detached from reality.

Productivity Concerns & Revenue Streams

The initial justification for the surge in AI spending was the expectation of rapid productivity gains. However, since the mainstream release of ChatGPT in November 2022, US labor productivity growth has only averaged around 2%, a slight increase from pre-ChatGPT levels. Corporate data corroborates this, with only 12% of companies reporting increased revenues and cost efficiency from AI adoption, while 55% report no material benefit.

Currently, the primary revenue driver within the AI space is large language models (LLMs) like ChatGPT, boasting roughly 1 billion monthly active users – the fastest adoption rate in history. However, LLM revenues (estimated at $300-$500 billion annually by 2030) represent only 7.5% of the projected $7 trillion in infrastructure spending. To justify the $7 trillion Capex, AI would need to generate a multi-trillion dollar revenue stream or deliver substantial, sustained productivity gains. Currently, there is no concrete evidence of either.

Potential Economic Impact & Future Outlook

The potential economic impact of a correction in AI spending depends on when the unwind occurs. A 30% pullback from the $375 billion spent in 2025 would equate to a 0.3% hit to GDP, even accounting for the multiplier effect. However, the longer AI spending continues to grow and exceed expectations, the more vulnerable the economy becomes to a significant downturn when investors realize the return on investment may not materialize.

The analysis suggests that 2026 may present further opportunities in semiconductor stocks as euphoria continues. The speaker’s firm has already profited from positions in companies like KAC, TSM, and ACMR and has a watchlist of stocks expected to outperform. They offer a free report detailing their investment strategy (link in the video description).

Notable Quote:

“The real risks to the economy are not the stocks themselves. It is all the elements tied to these stocks.” – Speaker

Technical Terms:

• Capex (Capital Expenditure): Funds used by a company to acquire, upgrade, and maintain physical assets such as property, plants, buildings, and equipment.
• GDP (Gross Domestic Product): The total monetary or market value of all final goods and services produced within a country’s borders in a specific time period.
• TSM (Taiwan Semiconductor Manufacturing): The world’s largest dedicated independent semiconductor foundry.
• KAC (Keyence Corporation): A Japanese manufacturer of sensors, measurement systems, laser markers, microscopes, and other factory automation equipment.
• ACMR (ACM Research, Inc.): A global leader in wafer processing equipment used by semiconductor manufacturers.

Conclusion:

The video presents a nuanced perspective on the current market, arguing that while a broad tech bubble may not be present, a potential bubble exists within the semiconductor sector driven by AI infrastructure spending. The core concern isn’t the stocks themselves, but the unrealistic expectations surrounding AI’s productivity gains and revenue potential. The analysis highlights the critical role of Taiwan in the global semiconductor supply chain and suggests that a decline in chip exports could significantly impact US economic growth. The speaker anticipates continued opportunities in semiconductor stocks in the near term but warns of increasing vulnerability as AI spending continues to grow without demonstrable returns on investment.