Lotus Microsystems has announced the launch of its vStrata power-delivery architecture, designed to enhance power efficiency in data centers. This initiative becomes increasingly crucial given the rising energy demands in both AI and non-AI environments.
At the core of the vStrata platform is the proprietary Power Interposer Technology (PIT), which optimizes power conversion by positioning it closer to the processor. This technology harnesses a vertical power delivery (VPD) chip, aiming to minimize electrical resistance by allowing power to be delivered directly through the processor’s package.
The vStrata architecture shortens current pathways and integrates thermal management directly into the power delivery system, helping to lower conversion losses and improve cooling efficiency. According to Lotus Microsystems’ claims, the system can achieve point-of-load efficiencies of up to 96%, resulting in over a 50% reduction in power conversion losses compared to traditional models.
CEO Hans Hasselby-Andersen emphasized the importance of a more efficient topology, which allows for more usable power from the input. He noted that their silicon substrate technology also addresses thermal issues associated with power conversion, thus further enhancing energy efficiency.
Despite the typical efficiency of power converters hovering around 90%, Lotus’s PIT aims for a 96% efficiency rate. This considerable improvement could have significant implications for data centers, particularly those employing multiple power-hungry GPUs, potentially reducing the reliance on water-intensive cooling solutions.
Lotus Microsystems claims that vStrata modules will be compatible with existing power-management systems, simplifying the integration process for semiconductor and system vendors. While the new power supplies are designed to advance data center technology, they are not retrofittable into existing server racks due to the absence of a standard footprint for server power supplies.
The first vStrata platform module, known as the LSC0580, is expected to ship as an engineering sample in the third quarter of 2026, signaling an important step towards improving energy efficiency in data center operations.
For those in the data center industry, this development underlines the growing focus on energy efficiency as a vital consideration in future infrastructure planning and deployment strategies.
Lotus Microsystems | Energy Efficiency | Data Center | Artificial Intelligence
