The AI bull market faces a critical fork in the road, with the semiconductor market returning to the debate of "1995 or 2000"

Mars Finance News, June 23 — The core battleground of AI trading is shifting from large technology stocks to semiconductors, but this uptrend is also beginning to show signs of the historic frenzy. The Philadelphia Semiconductor Index (SOX) is still moving within a steep upward channel, and a strategy of buying on pullbacks to the 21-day moving average has continued to work effectively this year. However, this trade is becoming increasingly crowded.

SOX is currently about 23% above the 50-day moving average. While it has not reached the extreme levels seen at the May stage high, short-term overbought conditions are already quite evident. Even more worth watching is that SOX’s monthly RSI has risen to near the high levels seen during the internet bubble period. This shows that the semiconductor trend remains strong, but momentum has entered a range that usually appears only in historical bouts of exuberance.

Money flows are also changing. The ratio of SOX to the Magnificent 7 has risen to its highest level since 2019, indicating that investors are replacing large technology stocks with semiconductors to express the AI theme. Goldman Sachs data also shows that the Magnificent 7’s net exposure has recently declined, suggesting that these tech giants are becoming a “source of funds” for chasing AI gains.

The volatility market is sending more complex signals. The VXN/VIX ratio has recently jumped sharply, indicating that technology stock volatility is rising faster than volatility across the broader market. The Market Ear believes that this combination of “spot prices rising while volatility is also rising” is unusual, meaning the trend is still strong, but the structure is becoming more fragile both upward and downward.

If we look at 1995 for reference, SOX also experienced a furious rally that year, followed by a painful correction, but that did not end the bull market; the true frenzy phase did not unfold until late 1998. In other words, today’s semiconductor rally may only be early overheating within a larger cycle.

But if we compare to 2000, the risk is higher. Comparing the MSCI Global Semiconductor Equipment Index with the Nasdaq’s performance from 1996 to 2003 shows that the current path of the semiconductor equipment sector has similarities to the late stage of the dot-com bubble. The author does not provide a definitive conclusion, but leaves the judgment to the market: the current trend has both the shadow of a bull market continuation and the outline of a late-stage bubble.

Speculative fervor in the Korean market has further intensified these concerns. On days with large swings, the scale of gamma rebalancing by traders in leveraged and inverse ETFs in Korea may exceed 20% of KOSPI’s daily trading value, meaning that leveraged products themselves could amplify market gains and losses.

Meanwhile, a rare divergence has also emerged between stock market volatility and interest rate volatility. A sharp drop in bond volatility is usually favorable for a rise in the stock market, but the S&P 500 has not fully reflected this signal. For bulls, this could mean there is still room for further upside; for bears, it suggests that the market’s current pricing of risk is not sufficient.