Chip stocks, as measured by the Philadelphia semiconductor index (SOX), are experiencing one of their worst months ever, down over 22%, driven by concerns over AI financing, international competition, and the sustainability of the investment cycle. While some analysts fear a deeper correction, portfolio managers like Dan Niles view it as a “speed bump” in a long-term uptrend, citing historical precedents of significant drawdowns during booms.
Key issues include a massive circular financing deal between Nvidia and OpenAI, increased competition from China in chip manufacturing, and a growing glut of AI-related debt, leading experts to question the pace of the AI boom’s continuation.
Chip stocks are experiencing significant pressure, driven by concerns over circular financing, intense international competition, and the sustainability of the current investment cycle in artificial intelligence. The Philadelphia semiconductor index (SOX) has plummeted over 22% this month through midday, marking its sixth-worst monthly performance ever.
Despite this sharp downturn, it remains uncertain whether this signals the end of the lucrative AI super-cycle. Historical data from the 1990s offers perspective: prior to its March 2000 peak during the dot-com bubble, the SOX index endured several months of substantial percentage declines comparable to the current one. This illustrates the difficulty in definitively calling a top in an investment boom that extends beyond typical business cycles. Only after the dot-com bubble burst in early 2000 did the semiconductor index ultimately lose nearly 80% over the subsequent two-and-a-half years.
Dan Niles, a portfolio manager at Niles Investment Management, characterized the current slide in chip stocks as a mere "speed bump" within a larger uptrend, albeit one that feels devastating. "These speed bumps can get to be really vicious. You go back to 1995 [and] semis were down over 50% … You had another drawdown of about 50% starting in late 1997. The semi index finished up 850% from the end of 1994 to its peak, so the speed bumps look like the end of the world until they're not," Niles explained to CNBC on Monday.
The AI trade is currently being scrutinized due to a massive circular financing arrangement between chipmaker Nvidia and AI platform and researcher OpenAI. Under a reported $250 billion deal, Nvidia would co-sign debt for OpenAI, enabling the latter to lease space in a new $500 billion data center in southern Ohio, currently under construction by a Softbank energy subsidiary. This deal, first reported by The Wall Street Journal on Sunday, mirrors a similar credit-wrapping structure employed by Alphabet to guarantee data centers used by software developer Anthropic, which, in turn, procures Alphabet's specialized AI hardware, tensor processing units.
However, some analysts interpret Monday's semiconductor stock slide, following news of the backstopping deal, as an indication that the current AI correction has further to fall. Jonathan Krinsky at BTIG advised clients that such a trade, "in the face of a well publicized 'backstop,' suggests we still have a ways to go in this AI correction." He forecasts a potential test of the 200-day moving average for the SOXX, an iShares ETF mirroring the semiconductor index.
Geopolitical competition with China is also casting a shadow over the AI sector. Shanghai Aishengna Electronic Technology Group has commenced manufacturing deep-ultraviolet photolithography machines, critical for advanced chipmaking, as The Information reported Monday. This development impacted Dutch semiconductor equipment giant ASML. Concurrently, Moonshot AI launched its Kimi K3, an open-weight AI software that offers a lower-cost alternative to products from Anthropic and OpenAI. Growing cost concerns have prompted several companies to re-evaluate their AI usage, combining various software components to enhance efficiency and reduce reliance on single models or workflows.
Perhaps the most significant concern is the escalating volume of AI-related debt. Adam Crisafulli highlighted in his Vital Knowledge newsletter on Tuesday, "As capex swamps free cash flow, [and] capital markets exhibit a diminished appetite for AI paper … Many are wondering how the boom can continue at the present pace for an extended period." Traders at UBS echoed this sentiment, drawing connections between debt, capital investment, and return on invested capital. They noted in a Tuesday memo that "While the AI debt boom is still very much alive, it has started to show some signs of fatigue, as recent [investment grade] tech supply has faced weaker demand and underperformed in secondary" markets.
