The semiconductor industry is just beginning to climb out of a painful downturn, with early signs of a new upcycle taking shape. Demand is stabilizing in smartphones and PCs, memory prices are rebounding, and inventory overhangs are finally being cleared. Yet despite being in what is arguably the early innings of this recovery, a growing number of semiconductor stocks—excluding AI titans like Nvidia, AMD, Broadcom, and TSMC—appear to be priced as if they’re already deep into the boom phase. The AI gold rush has sparked a massive revaluation across the sector, lifting not only those at the core of the revolution, but also many on the periphery whose earnings have yet to reflect the optimism embedded in their share prices.
This dynamic is especially visible in companies that are AI-adjacent but not core beneficiaries. Marvell Technology, for example, has attracted substantial investor attention for its custom silicon solutions for cloud and networking infrastructure, yet a significant portion of its revenue still comes from non-AI applications where growth remains tepid. Nevertheless, the stock trades at premium multiples well above its historical norms. The same story unfolds with companies like Rambus, GlobalFoundries, and memory suppliers such as Micron and SK Hynix, where surging share prices reflect sky-high expectations for high-bandwidth memory demand, even though their traditional DRAM and NAND businesses are only just beginning to recover.
What’s driving this enthusiasm is the belief that AI will eventually ripple through every layer of the semiconductor ecosystem—touching memory, storage, networking, analog ICs, power management chips, and even foundry services at mature nodes. That vision may be directionally correct. But the market has gone one step further and priced many of these stocks as if that future is already here, or at the very least, guaranteed. It’s a rerating based not just on potential, but on assumed inevitability.
The problem is that fundamentals are still catching up. Most companies in these secondary or tertiary positions in the AI value chain are coming off weak earnings bases. The downturn in 2022 and 2023 decimated profitability, and even with early signs of stabilization, revenue growth across the board remains modest outside the AI core. As a result, valuation multiples are not only elevated because of bullish sentiment—they’re also artificially inflated by depressed earnings in the denominator. Price-to-earnings and price-to-sales ratios appear stretched, not because performance is robust, but because the bar for future growth has been set uncomfortably high.
This doesn’t mean these companies are doomed to disappoint. If AI adoption broadens significantly, if the cyclical recovery continues smoothly across industrial, automotive, and consumer segments, and if capital spending remains robust, then some of these stocks may grow into their valuations. But the margin for error has become razor thin. Any delay in demand recovery, any hiccup in AI capital expenditures, or any geopolitical friction impacting the supply chain could trigger a reassessment.
Ironically, while the semiconductor sector is cyclical by nature, much of it is now priced as if the cycle has already reached a midpoint. That leaves less room for upside and more sensitivity to macro or execution risks. Investors who assume that all chipmakers will ride Nvidia’s coattails may be overlooking the nuances of where value is actually being created—and how much of that value is already assumed in today’s stock prices.
So while the broader chip industry is just starting its next cycle, many of its constituents have already seen their shares rerated as if the boom is in full swing. This doesn’t necessarily portend a crash, but it does demand caution. For investors, the challenge now is separating the truly transformative players from those merely riding the AI wave. Because in this early stage of the upcycle, prices in many corners of the semiconductor market already look dangerously late.