Over the past two years, the keyword of the global semiconductor market has been almost singular: AI compute. From advanced-node foundry and HBM high-bandwidth memory to optical modules and advanced packaging, capital has flowed along the compute industry chain, driving up valuations of related assets and giving the rally a rare "one-sector heat" pattern. However, entering Q4 2026, the market's micro-thermometer is showing subtle changes—as crowding in AI-themed trades reaches historic highs, a growing number of industry and investment perspectives are asking: could the "second half" of the semiconductor rally be carried by power semiconductors and analog chips? From three dimensions—supply-demand structure, price logic, and industry chain transmission—this article sorts out the underlying logic of this potential sector rotation.
1. Why Sector Rotation Always Brews at the "Hottest Moment"
Looking back at past semiconductor cycles, market broadening has been almost an inevitable script: cycles usually start from the segments with the tightest supply and the greatest technological change, then transmit step by step along the chain of "advanced nodes—mature nodes—analog and power—equipment and materials." When price hikes and capacity expansion in leading segments enter their delivery phase, marginal capital automatically seeks the intersection of valuation troughs and fundamental inflection points.
The current market structure happens to meet these conditions. On one hand, capacity bottlenecks in the AI compute chain are being gradually eased by massive capital expenditure, and declining supply-side scarcity means excess profits will eventually revert to the mean; on the other hand, power and analog chips have gone through two years of destocking and price adjustment, the industry shakeout is nearing its end, and fundamentals sit in a typical left-side zone where bad news is losing bite while good news has yet to materialize. This scissors gap between valuations and industry sentiment is precisely the soil in which rotation most easily takes root.
2. Power Semiconductors: Out of the Price War as Demand Diversifies
The core contradiction in power semiconductors' last downturn lay on the supply side: silicon carbide substrate and device capacity was released in concentration over recent years, compounded by EV growth downshifting; oversupply triggered fierce price competition, product prices fell rapidly, and industry margins were broadly squeezed. But at the current point, three marginal changes deserve close attention.
- Diversifying demand structure. Power supply architectures for AI data centers are evolving toward higher voltage levels, with new solutions such as high-voltage DC power and solid-state transformers sharply raising the quantity and specification requirements for power devices; meanwhile, emerging scenarios such as integrated solar-storage-charging, humanoid robots, and the low-altitude economy are opening a "second growth curve" for power chips beyond EVs.
- The industry shakeout is nearing its end. The prolonged price war has forced some capacity lacking cost advantages and technological depth to exit or transform, utilization in the substrate segment has diverged markedly, and leading players' market share and pricing power are set for systematic repair once the shakeout is complete.
- Technology premiums are being repriced. Automotive-grade certification, substrate-device integration capability, and accumulated packaging know-how are again becoming the dividing line; players with full-chain capabilities will shift from "trading price for volume" to "pricing on quality."
3. Analog Chips: A "Deep Squat, Slow Rise" Cycle, with Elasticity from Product Portfolios
Unlike digital chips, analog chips have long life cycles, low per-unit value, but highly dispersed part numbers; their cycle features "slow falls and slow rises," yet once recovery sets in it tends to be durable. Current fundamental signals in the analog industry show typical bottom characteristics: consumer demand is recovering first, industrial and automotive customers are shifting from active destocking to on-demand restocking, and channel inventory is gradually falling back to healthy levels.
More crucially, the recovery elasticity of analog chips depends heavily on portfolio structure. Platform players with rich part numbers close to industrial and automotive long-tail customers can enjoy multi-end-market resonance sooner, while players concentrated in a single consumer scenario must wait for clearer end-demand recovery signals. This structural divergence means notable differentiation will emerge within the analog sector itself, making stock selection more important than overall positioning.
4. Transmission Up the Chain: Opportunities in Mature Nodes and Materials
The recovery in power and analog chips will map directly onto wafer foundry and upstream materials. Mature-node capacity utilization is a key leading indicator for this rotation—when utilization at 8-inch and 12-inch mature nodes keeps rising, foundry pricing power will improve accordingly, and the earlier passive stance of "securing prices with volume" may be reversed. Meanwhile, demand for photoresists, wet electronic chemicals, lead frames, and packaging substrates will expand in step, and upstream "shovel sellers" often shine during the market-broadening phase.
5. Implications for the SGX Tech Segment and Industry Decision-Makers
Singapore is a vital hub in the global semiconductor industry chain, hosting core assets across equipment, assembly and test, electronics manufacturing services, and materials, and it is highly sensitive to the rally broadening from AI into power and analog. For investors, rather than continuing to chase marginal moves in the crowded AI theme, it may be better to focus early on power and analog chains whose fundamentals sit at the bottom with improving supply structures, as well as the linked mature-node foundry and materials assets.
For procurement and supply chain decision-makers, power and analog chip prices are at the cycle bottom, a window for locking in medium- to long-term supply agreements and optimizing cost structures. Historical experience shows that long-term contracts signed at the bottom often deliver significant cost advantages and supply security in the next upcycle.
Of course, rotation will not necessarily arrive on schedule. If the AI capex boom lasts longer than expected, capital may stay with the compute theme; if the power industry's price war drags on beyond expectations, the shakeout will slow. Investors and buyers should track three signals: the direction of change in mature-node utilization, the month-over-month trend in power device quotes, and inventory turnover days at analog chip makers.
Conclusion
The semiconductor market has never been a single-voice chorus but a polyphonic symphony of rotation. As the AI compute main theme reaches its climax, the repair of the "bass voices" in power and analog may well be the prelude to the next movement. For investors and industry decision-makers positioning for 2027, the most worthwhile task now is to study carefully the supply-demand statements that are quietly improving before market consensus forms—because every market broadening in history began with a few people seeing the inflection point first.
