On June 12, 2026, Goldman Sachs raised its price target for Infineon Technologies from €75 to €88, maintaining a 'Buy' rating. This move appears to be a routine valuation adjustment for a European semiconductor firm—but it signals something deeper: the expansion of AI infrastructure is recalibrating how markets value “non-compute” semiconductors. While investors fixate on GPUs, HBM, and advanced nodes, they overlook the silent enablers of the entire AI stack—power management and power conversion chips. Infineon happens to be the global leader in precisely this domain.
Infineon is not a traditional AI chipmaker. It doesn’t produce GPUs for training large language models, nor does it manufacture high-bandwidth memory. Yet its products—IGBTs, SiC MOSFETs, and power management ICs—are indispensable in data centers, AI servers, and even edge AI devices. According to its fiscal 2025 report, revenue from its Power & Sensor Systems (PSS) segment grew 21% year-over-year, with data center–related sales now exceeding 30% of the segment total. That share has doubled over the past three years, far outpacing consensus expectations.
Goldman Sachs analyst Alexander Duval highlighted that AI server power density is increasing at roughly 30% annually. A single server equipped with eight NVIDIA Blackwell GPUs can draw over 15 kilowatts—more than triple the consumption of comparable systems five years ago. This exponential rise places unprecedented demands on power efficiency, thermal management, and energy conversion. Infineon’s CoolSiC™ and OptiMOS™ technologies deliver stable operation at conversion efficiencies above 98%, making them the preferred choice for leading cloud providers and server OEMs. Microsoft Azure and Amazon AWS have both deployed Infineon power modules at scale in their next-generation AI infrastructure—a trend unlikely to reverse.
More importantly, Infineon’s business model carries inherent cyclicality resistance. Unlike logic chipmakers locked in an arms race over nanometer nodes, power semiconductors rely more on materials innovation (e.g., silicon carbide), packaging integration, and system-level design. These capabilities are hard to replicate quickly, and customer switching costs are extremely high. Once embedded in a Tier 1 supply chain, relationships often endure for a decade or more. This “embedded stickiness” ensures resilience during downturns: in 2024, when global semiconductor sales declined by 5%, Infineon maintained an operating margin of 19.3%, well above the industry average of 14.7%.
Bernstein Research recently reinforced this thesis. Its analysis estimates that demand for high-efficiency power management chips in AI data centers will grow at a 24% CAGR through 2028—far outpacing the broader semiconductor market’s projected 8%. Infineon’s decades-long expertise in automotive and industrial power systems has enabled it to build a full-stack capability spanning materials, devices, modules, and system-level solutions. This vertical integration grants it greater flexibility in responding to AI workload volatility and tightening energy regulations, such as the EU’s Ecodesign for Sustainable Products Regulation (ESPR) effective in 2027.
Yet the market continues to price Infineon through the outdated lens of an “automotive electronics company.” While its auto segment indeed contributes nearly half of revenue, the growth trajectory of its AI-related business is markedly steeper. Trading at around 18x forward P/E, Infineon trades at a discount to TSMC (22x) and ASML (25x)—and far below pure-play AI chip firms. This mispricing presents a structural opportunity.
I judge that over the next 18 months, as AI infrastructure shifts from “compute-first” to “efficiency-first,” the strategic importance of power semiconductors will become undeniable. Infineon stands to benefit not only from AI server power demands but also from emerging applications in AI PCs, edge inference devices, and smart grids. Its competition with STMicroelectronics and onsemi is no longer just about specs—it’s evolving into an ecosystem battle over holistic energy-efficiency solutions.
One question lingers: as the industry obsesses over FLOPS per watt, who ensures that every kilowatt-hour actually becomes useful computation? Infineon’s answer may be worth more than we think.