Summary of Key Points
The AI and chip sectors, which drove the U.S. stock market upward in the first half of the year, have recently suffered a significant decline. Over 60% of technology stocks have entered a “bear market” (with declines of more than 20%), but there is ongoing debate about whether this drop is a short-term correction or a revaluation of valuations. Most analysts believe it is a normal profit-taking after the previous sharp gains, while Goldman Sachs insists that the AI-driven tech cycle has not yet peaked and continues to be optimistic about related sectors.
How Bad Have Technology Stocks Fallen?
In the S&P 500 Information Technologies sector, 69% of stocks have fallen by more than 20% from their recent annual highs (this is the commonly recognized criterion for a “bear market”). Semiconductor leaders have performed particularly poorly: Micron Technology has dropped by 25%, Broadcom by 21%, and Marvell Technology by 30%. The storage sector has seen even more dramatic declines, with a sharp turnaround from rapid gains in the first half of the year to a sudden plunge at the end of June, resembling a rollercoaster ride.
Why the Sudden Sharp Drop?
1. Profit-Taking: Stocks rose too much, and investors are looking to lock in their profits. The Philadelphia Semiconductor Index saw unprecedented growth in the second quarter. After earnings reports were released, investors sold their positions (a professional term for “profit-taking”). For example, technology infrastructure stocks collectively declined on Tuesday. This pattern has been consistent over the past few quarters: technology stocks usually fall within a month after earnings reports until they rebound before the next reporting period.
2. Samsung’s Earnings: Samsung just announced preliminary earnings showing a 19-fold increase in operating profit year-over-year, but investors had higher expectations, leading to concerns that chip prices for manufacturers like Micron might slow down. Since storage chip prices have been rising, investors are selling their positions.
3. Cycle Anxiety in the Storage Industry: The storage industry is highly cyclical, with price fluctuations depending on supply and demand. There is debate about whether AI demand will fundamentally change this cycle. For instance, AI requires a large number of storage chips; could this lead to sustained demand and break the previous pattern of price swings?
What Are Analysts Arguing About?
1. **“Normal Correction Camp”: This is just a temporary pause after significant gains. Morningstar analysts believe it’s a normal recovery process and there’s no need for panic. Evercore analysts also think the storage sector remains valuable, and current reductions in holdings are reasonable profit-taking actions as the market re-evaluates how long chip price increases will last and whether cloud companies will spend less on hardware.
2. “Cycle Concern Camp”: Similar to the 2000 commodity boom: NS Partners partners argue that although technology stocks have risen significantly, valuations have decreased, suggesting they may not be in a bubble. However, the market structure resembles the early 21st-century commodity supercycle, where demand surged but supply couldn’t keep up, leading to a sharp drop once supply increased.
3. **“AI Profit Concern Camp”: Hardware investment depends on revenue generated by AI applications. Moore’s Law warns that while there is strong demand for hardware (such as in storage and liquid cooling), it ultimately relies on AI applications (like ChatGPT) to generate sufficient revenue to justify the investment. If applications don’t generate enough profit, demand for hardware will eventually decline.
Why Is Goldman Sachs Still Optimistic?
Goldman Sachs believes the AI-driven tech cycle has not peaked for two reasons:
1. No Signs of Excess Supply or Technological Slowdown: The two indicators of a cycle ending are an oversupply of semiconductors and a slowdown in technological innovation, with companies shifting to price competition rather than performance competition. Neither of these signs is present yet.
2. Future Demand: Investments in AI infrastructure are still expanding, and future trends like physical AI and edge AI (in phones, cars, etc.) will drive continued hardware demand, prolonging the cycle.
Goldman Sachs’s investment recommendations include:
- Continuing to buy stocks related to AI servers and data centers;
- Selecting individual stocks carefully in sectors with tight supply and demand;
- Buying software/IT services companies that use AI to expand their businesses (as they are more stable).
What Risks Should Be Watched Out For?
1. Retailers Using Leverage: Edward Jones strategists point out that retail investors using leveraged ETFs to buy AI stocks indicates speculative overheating, which could amplify any potential declines.
2. Supply Catching Up with Demand: If new production capacity is built or new companies enter the AI hardware market, supply will increase, reducing companies’ pricing power and potentially leading to profit declines, marking the mid-to-late stages of a cycle.
3. Slowing AI Capital Spending: Even if AI demand continues to grow, if cloud companies (like Amazon, Microsoft) reduce their spending on hardware, related stocks could experience significant drops.
In summary, this decline in technology stocks is more like a “mid-game break,” but the key factors are whether AI applications can generate sufficient revenue and whether supply will exceed demand. If these issues remain unresolved, the AI cycle may continue; otherwise, caution is advised. Goldman Sachs’s optimism is not unfounded, but risks must also be taken seriously.