人工智能:Soitec 和意法半导体(STMicroelectronics)两大巨头准备降低数据中心的能耗

IA : les géants Soitec et STMicroelectronics prêts à réduire la consommation électrique des data centers

La Tribune AURA Original
摘要
尽管业绩依然低迷,分析师对两家半导体制造商仍持乐观态度,认为它们此前因错过AI处理器机遇而受批评的局面可能即将扭转。

尽管Soitec与STMicroelectronics两家半导体巨头的财务表现持续低迷,分析师此前也批评它们未能及时抓住人工智能处理器的市场机遇,但近期看法已明显转向乐观。背后的关键因素在于,这两家公司正致力于通过创新技术大幅降低数据中心的电力消耗——这恰好切中当前AI算力爆发式增长带来的能耗痛点。随着能效优势逐渐显现,它们有望在AI基础设施领域重新夺回主动权,逆转此前被动的市场处境。

Summary
Analysts remain optimistic about two semiconductor manufacturers despite ongoing weak results, as they may be turning the corner after previously facing criticism for lagging in AI processor development.

Soitec and STMicroelectronics, two French semiconductor heavyweights, continue to post lackluster financial results. Yet industry analysts covering both companies are striking an increasingly optimistic tone. The reversal in sentiment stems from a strategic pivot long demanded by investors: both firms are finally positioning themselves to capitalize on the artificial intelligence boom—not by designing AI processors themselves, but by tackling one of the sector’s most pressing physical constraints, the soaring electricity consumption of data centers.

For several quarters, Soitec and STMicroelectronics were criticized for missing the AI wave, focusing instead on traditional automotive, industrial, and communications end markets. That narrative is now shifting. Executives at both companies have signaled that their advanced materials and chip technologies can play a central role in improving energy efficiency across the AI infrastructure stack. Soitec’s engineered substrates, particularly its silicon-on-insulator (SOI) wafers, enable transistors to switch faster and leak less current, directly cutting power usage in high-performance computing environments. STMicroelectronics, meanwhile, is leveraging its expertise in wide-bandgap semiconductors—such as silicon carbide (SiC) and gallium nitride (GaN)—as well as energy-efficient microcontrollers and power management ICs, all of which are critical to reducing losses in server power supplies and voltage regulators.

The addressable opportunity is enormous. Data center power demand is projected to double by 2030, fueled by the exponential growth of AI training and inference workloads. Hyperscalers and enterprises alike are under pressure to decouple compute growth from energy consumption, creating a fast-growing market for components that can deliver performance without proportional power increases. Even if Soitec and STMicro are not directly selling AI accelerators, their technologies are becoming essential enablers behind more sustainable data center design.

While revenue and margin figures remain under pressure from inventory corrections and softening demand in legacy segments, the strategic repositioning is giving analysts confidence that these companies are finally aligned with long-term secular trends. If the promised energy-efficiency gains translate into design wins and volume orders, the current financial trough could mark a turning point. The wind, it seems, is at last blowing in their direction.

Résumé
Malgré des résultats financiers toujours décevants, les analystes se montrent optimistes pour deux grands fabricants de semi-conducteurs, jusqu’ici critiqués pour leur retard dans les processeurs dédiés à l’IA. Un possible retournement de tendance dans ce secteur stratégique pourrait améliorer leur position concurrentielle.

Malgré des résultats toujours en berne, les analystes des deux fabricants de semi-conducteurs restent optimistes. Car si ces derniers étaient épinglés pour ne pas avoir pris le virage des processeurs IA, le vent pourrait bien être en train de tourner.

AI Insight
Core Point

Soitec and STMicroelectronics are pivoting toward AI processors that reduce data center power consumption, potentially reversing their earlier failure to capitalize on the AI chip boom.

Key Players
  • Soitec — semiconductor materials supplier, based in Bernin, France.
  • STMicroelectronics — semiconductor manufacturer, headquartered in Geneva, Switzerland.
Industry Impact
  • ICT: High — power-efficient AI chips directly lower data center operational costs and energy use.
  • Computing/AI: Medium — addresses growing demand for sustainable high-performance processors.
  • Energy: Medium — reduces electricity strain and carbon footprint of expanding data centers.
Tracking

Monitor — early pivot signals a turnaround; watch for design wins and market adoption.

Related Companies

No companies linked yet

Categories
半导体 人工智能
AI Processing
2026-06-04 12:38
deepseek / deepseek-v4-pro