Infineon targets AI power density with TLVR quad-phase module

Infineon Technologies has introduced what it describes as the industry’s first TLVR-based quad-phase power module exceeding 2 A/mm², aimed squarely at next-generation AI compute platforms. The new TDM24745T OptiMOS module integrates multiple power components into a compact footprint to address rising power density demands in AI data centers, the company notes.

For eeNews Europe readers, the announcement highlights a key bottleneck in AI infrastructure — efficient power delivery — and offers insight into how semiconductor vendors are tackling scaling challenges at the system level.

According to the release, the TDM24745T integrates four power stages, a trans-inductor voltage regulator (TLVR) inductor, and decoupling capacitors within a 9 × 10 × 5 mm³ package. This high level of integration enables a current density exceeding 2 A/mm² while supporting both lateral and vertical power delivery architectures.

“As AI workloads scale at unprecedented speed, the need for highly efficient and extremely compact power delivery has never been greater,” said Athar Zaidi, Senior Vice President and General Manager Power ICs and Connectivity at Infineon, in the release. “With the TDM24745T, we are redefining what is possible in high-current voltage regulation. By combining industry-leading current density with TLVR technology in an exceptionally small footprint, we enable our customers to unlock more compute performance, reduce energy consumption, and accelerate the deployment of next-generation AI data centers.”

The module delivers up to 320 A peak current capability, positioning it for high-current multiprocessor platforms and advanced AI accelerators.

Infineon says the TLVR architecture enables faster transient response while reducing required output capacitance by up to 50%, simplifying system design. This reduction can translate directly into PCB space savings, allowing designers to allocate more area to compute resources.

The company also highlights improved thermal and efficiency performance through the use of OptiMOS-6 MOSFET technology, chip-embedded integration, and proprietary magnetics. Combined with Infineon’s digital multiphase controllers, the module supports scalable and flexible power architectures suited to rapidly evolving AI workloads. The TDM24745T fits into Infineon’s broader AI power ecosystem, which spans grid-to-core power delivery and leverages silicon, silicon carbide (SiC), and gallium nitride (GaN) technologies to optimize efficiency and robustness in data center designs, the company notes.

As AI infrastructure continues to scale, power delivery is becoming a defining constraint. By integrating multiple functions into a compact module and improving transient response, Infineon aims to reduce system complexity while improving total cost of ownership (TCO), according to the release. For system designers, the combination of higher current density, reduced passive component requirements, and improved efficiency could help accelerate deployment timelines and enable denser compute architectures.