Following a blockbuster second quarter, chip stocks started the third quarter with a sharp downturn. The VanEck Semiconductor ETF (SMH), which tracks chip stocks, fell over 5% on Wednesday, after posting its best quarter ever with a 71% jump from April to June. This sell-off was widespread, with Micron falling 11%, Intel down 9%, and Advanced Micro Devices (AMD) dropping 7%. Equipment manufacturers like Lam Research, KLA Corp., and Applied Materials also saw declines of at least 10%.
A primary catalyst for the market volatility was a Bloomberg report indicating that Meta Platforms is considering launching a cloud business to rent out its excess AI computing capacity. This news fueled fears that the supply of AI processing might be catching up to demand, potentially signaling a slowdown in the massive AI spending boom by tech giants. JPMorgan analyst Nikolaos Panigirtzoglou suggested that the steady outperformance of semiconductor stocks over hyperscalers since last September might be unsustainable in the long run.
The decline extended into Thursday's premarket trading, with Micron and Western Digital falling about 2.1% each, and Coherent and Marvell Technology shedding around 2% and 1.8% respectively. AMD, Intel, and Microchip Technology also saw about a 1% dip. In Asian markets, South Korea's Kospi index tumbled 8%, primarily driven by steep losses in chipmakers Samsung Electronics and SK Hynix. SK Hynix dropped 14.57%, and Samsung shed 9.06%.
Despite the sell-off, Meta Platforms itself bucked the trend, rallying 8.8% on the day the news broke, as investors welcomed the potential to generate revenue from its excess capacity. Meta, which plans to spend up to $145 billion on capital expenditures this year for data centers and GPUs, could leverage its AI infrastructure. Analysts from KeyBanc Capital Markets view Meta's move as positive, positioning the company further into the enterprise market and potentially offering a quicker return on investment. Some analysts believe the market reaction to the news was excessive, arguing that these developments don't necessarily signal a significant slowdown in AI infrastructure investment or a material deterioration in fundamentals.
