Watchlist

What's going on

The drop is a multi-week "growth stock" de-rating in AI/semiconductor names plus an HSBC downgrade to Hold on July 14 citing foundry capacity bottlenecks that could cap near-term earnings, not any fraud, guidance cut, or lost customer.

The case for it

Bulls would say this is just profit-taking after the stock more than doubled off its 52-week low, with revenue and earnings still growing 20%+ and 48% respectively, a debt-free balance sheet, and analysts still rating it a consensus Buy. But at a trailing P/E of ~335x and forward P/E of ~91x, the stock still requires years of flawless, hyperscaler-driven "AGI CPU" execution to justify the price — it isn't statistically cheap by any normal measure, it's merely cheaper than an extreme bubble price.

What could go wrong

Even after the crash the valuation leaves almost no margin of safety: <cite index="19-18">management targets $9 non-GAAP EPS by FY31, and shares still trade at an expensive 139x/95x 2026e/2027e earnings</cite>. On top of that, foundry supply constraints could delay the AI CPU revenue hyperscalers are counting on, RISC-V is emerging as an open-source competitive threat, and a Qualcomm litigation/China-exposure overhang adds tail risk into the July 29 earnings report — any stumble at these multiples could mean another 30-50% downside.