AI Chip Stocks Drop $2.3T Despite Bullish Analyst Targets

Artificial intelligence semiconductor stocks have shed more than $2.3 trillion in value since late June, according to Yahoo Finance, even as Wall Street analysts continue raising earnings forecasts and price targets across the sector. The disconnect between analyst optimism and investor behavior is creating what one industry observer calls a "wall of worry" for AI stocks.

Why it matters

This divergence has occurred only three times in the past 25 years—during the dot-com crash, the 2021 tech peak, and the 2018 semiconductor downturn. In two of those cases, stocks remained significantly lower six and twelve months after the peak. The pattern suggests investors may be pricing in risks that analysts' models don't fully capture, including demand saturation, competitive disruption, or capital expenditure pullbacks from hyperscalers.

Analysts maintain aggressive targets

Analyst projections remain bullish despite the sell-off, and they're not relying on inflated valuation multiples. KeyBanc's John Vinh raised his Micron price target to $1,750 from $1,600 following a trip to Asia, citing memory shortages expected to persist for at least 18 months. That target implies a 93% gain from Micron's current price of $904, based on a conservative nine-times price-to-earnings ratio applied to fiscal 2027 earnings.

Micron has fallen more than 25% since June 30 alone.

A survey of over 50 analyst updates across the past 60 days covering AI accelerators (Nvidia, AMD), custom silicon (Broadcom, Marvell, Arm), AI servers (Super Micro Computer), memory (Micron), foundries (TSMC), and CPUs (Intel, Qualcomm) shows widespread target increases. The largest gaps between analyst targets and current prices appear in companies that either experienced rapid gains—like Micron and Arm—or have long traded below analyst expectations, notably Nvidia.

Investor fear outweighs optimism

Investors sitting on substantial AI gains are protecting profits amid several concerns. Questions center on when demand will peak, whether competitors can erode market leaders' positions, and if capital spending by cloud providers will plateau or decline. The brief panic surrounding DeepSeek's emergence earlier this year demonstrated how quickly competitive fears can move markets.

Every earnings report now faces intense scrutiny for signs of weakness. TSMC's upcoming earnings release represents one such test, according to Karl Freund, founder and principal analyst at Cambrian-AI Research, who examined the divergence in a Forbes analysis.

Historical precedents offer caution

Freund identified three comparable periods of divergence between analyst targets and market performance over the past 25 years. During the dot-com peak and the 2021 tech bubble, stocks traded considerably lower six and twelve months after their peaks. The 2018 semiconductor and memory peak eventually recovered, but required a full year. While three data points don't constitute a definitive pattern, they suggest the current divergence warrants attention.

The analysis, which included clients of Cambrian-AI Research such as Nvidia and Qualcomm, was first reported by Freund at Forbes. He noted that while limited historical comparisons shouldn't drive investment decisions alone, they "give one pause" about the sector's near-term trajectory.

Investors must now weigh legitimate concerns about market saturation and competition against the transformative potential of AI infrastructure buildout—a tension Freund describes as markets climbing their characteristic "walls of worry."