Semiconductor Stocks Ride Big Tech’s AI Investment Boom as Downside Bets Mount Amid Profitability and Bubble Concerns

Single SMH Put Purchase Accounts for More Than 30% of Total Premiums
Big Tech’s Massive AI Spending Spree Raises Questions Over Monetization
Semiconductor Stocks Face Turbulence if the AI Bubble Bursts

Investors are increasingly betting on declines in semiconductor stocks. Big Tech companies are ramping up capital expenditures amid the artificial intelligence (AI) boom, fueling concerns that delays in monetizing their vast infrastructure investments could rapidly erode the semiconductor sector’s growth momentum. Some market participants, however, argue that it remains premature to adopt aggressive bearish strategies such as short selling, given the continued strength of AI demand and Big Tech’s investment drive.

Mounting Valuation Pressure on Semiconductor ETFs

According to a GuruFocus report published on August 19, a large put-option transaction worth $129 million was executed in the VanEck Semiconductor ETF (SMH), a leading global semiconductor exchange-traded fund, on August 18. The single put-option purchase accounted for more than 30% of total premiums in the massive ETF, whose market capitalization approaches $68.59 billion. GuruFocus reported that retail investors were gravitating toward call options in anticipation of further gains, while major institutional investors were using put options to hedge risk.

Valuation pressure appears to be driving the institution’s large bearish wager. According to GF Value, GuruFocus’ proprietary valuation model, SMH’s fair value—based on historical trading multiples and growth metrics—is estimated at $429.07. That figure is 32.2% below its current market price of $567.30. Its price-to-earnings ratio also stands at 24.26, above its historical median, while the valuation component of its composite GF Score is just 4 out of 10.

Big Tech’s Massive Capital Expenditures

The elevated valuations of semiconductor stocks are closely tied to conditions in the AI market. Major US technology companies have recently continued to pour vast sums into AI. Google, for example, said during its second-quarter earnings announcement last month that its capital expenditures this year would exceed its previous forecast. “We have significantly expanded capacity over the past three years, but demand continues to outpace our investment,” Google Chief Financial Officer Anat Ashkenazi said at the time. “We are raising our full-year capital expenditure guidance from $180 billion–$190 billion to $195 billion–$205 billion.”

Amazon also raised its full-year capital expenditure forecast during its second-quarter earnings announcement that month, lifting the estimate from $200 billion to $220 billion. The company is allocating most of its capital expenditures to expanding Amazon Web Services (AWS) and generative AI infrastructure, with spending reaching $53.1 billion in the second quarter alone. Microsoft (MS) plans to invest $175 billion in capital expenditures this year and spent $41 billion in the latest quarter alone. Meta has likewise projected annual capital expenditures of $130 billion–$145 billion. Its second-quarter capital expenditures totaled approximately $31.08 billion.

Uncertain Profitability of AI Businesses

The central question is whether AI profitability can keep pace with the scale of investment. AI demand remains robust for now. Cloud and AI-related revenue at major technology companies continues to rise, and some companies are struggling to meet demand without expanding their data centers. Such short-term strength, however, cannot guarantee long-term returns. Funds invested in graphics processing units (GPUs), servers and data centers will be recognized as depreciation expenses over several years, while revenue and profit generated by AI services could fluctuate sharply with industry conditions.

The AI industry is characterized by exceptionally rapid technological change. New GPUs and AI accelerators are continuously entering the market, while model-computing efficiency can improve within a short period. Existing infrastructure therefore faces a substantial risk of losing economic value sooner than anticipated. Intensifying competition among AI models could also drive down the prices of cloud computing and AI services, reducing the returns generated by AI infrastructure. This presents a substantial burden for Big Tech companies already grappling with cash-flow pressure from large-scale investment. Alphabet spent $44.9 billion on capital expenditures in the second quarter, pushing quarterly free cash flow into a deficit of $5.9 billion. Amazon also recorded negative free cash flow for a second consecutive quarter after capital expenditures exceeded operating cash flow, while MS’s gross margin continued to decline.

Risks Embedded in Semiconductor Stocks

Semiconductor companies are also expected to suffer a substantial blow if an AI bubble collapse materializes. The semiconductor industry has been the largest beneficiary of Big Tech’s AI investment boom. Regardless of whether AI businesses generate profits, Big Tech companies must first invest heavily in GPUs, high-bandwidth memory (HBM), servers, networking equipment and other infrastructure to develop AI models and deliver related services. If AI profitability fails to keep pace with rising costs, Big Tech companies are highly likely to slow new data-center construction and server expansion.

Such a shift would probably reverberate across the semiconductor supply chain. Slower data-center expansion would curb growth in GPU and HBM orders, potentially ending the prolonged period of supply shortages and price increases. If inventories subsequently accumulate and utilization rates decline, semiconductor companies’ profitability would deteriorate rapidly. Stock prices could absorb the shock before earnings begin to weaken. “Current semiconductor stock prices already reflect considerable expectations that AI capital expenditures will continue to increase over the next several years,” one market expert said. “Even a delay in Big Tech’s investment plans or a shift toward more conservative spending could simultaneously reset growth expectations and valuations across the semiconductor sector.”

Table 1. Conditions and Risks in the Global AI and Semiconductor Markets

Shifting AI and Semiconductor Investment Strategies

The recent rise in short bets against semiconductor and AI stocks stems from these risks. Michael Burry, the investor who rose to prominence for his “Big Short” trade, recently said that the size of his current short position was comparable to his bets during the 2008 financial crisis and the 2020 market crash. Burry has substantially increased his put-option position in the Invesco QQQ Trust (QQQ), an ETF concentrated in large technology stocks, while maintaining put positions in Nvidia and Palantir. Global hedge funds have also expanded short positions in AI-related stocks. According to Reuters, citing a report from data platform Hazeltree, hedge funds increased their short positions in AI server manufacturer Super Micro Computer, AI cloud infrastructure provider CoreWeave and AI infrastructure and cloud company Nebius Group. All three remained among the 10 most heavily shorted stocks last month after appearing on the list in June.

Some investors warn that indiscriminately increasing short exposure would be premature. Steve Eisman, the investor known from the film The Big Short, recently said on his podcast Weekly Wrap that there was still insufficient data to establish the risk of an AI market collapse. With hyperscaler investment and AI-related demand continuing to rise, Eisman reportedly maintains substantial long positions in AI-related assets. At the same time, he identified hyperscalers’ heavy dependence on OpenAI and Anthropic as the AI market’s “Achilles’ heel.” The two companies account for a significant share of AI and cloud revenue at MS, Amazon and Alphabet, leaving Big Tech directly exposed to any deterioration in their profitability. Eisman said such a scenario could prompt hyperscalers to cut capital expenditures, triggering a chain reaction that would weaken demand for AI infrastructure and semiconductors.