Texas Instruments introduced a complete 800V direct current (DC) power architecture for next-generation AI data centers, developed in collaboration with NVIDIA and based on the NVIDIA 800 VDC reference design. The solution addresses power challenges in AI infrastructure by maximizing conversion efficiency and power density across the entire power path, simplifying the architecture, and supporting more scalable and reliable operations in data centers.
TI's approach uses only two conversion stages from 800V to GPU core power: a compact 800V to 6V isolated bus converter, followed by a 6V to less than 1V multiphase buck solution with high current density compared to previous generations. This streamlined design aligns with the NVIDIA reference design.
The complete power solution includes multiple reference designs:
800V hot-swap controller for scalable input power protection on 800V rails.
800V to 6V DC/DC bus converter with integrated GaN power stages, achieving 97.6% peak efficiency and power density exceeding 2000W/in³ for compute tray applications.
6V to less than 1V multiphase buck converter for high-current delivery to advanced GPU cores, offering higher power density than 12V designs and featuring dual-phase power stages.
Additional showcased elements include:
A 30kW 800V high power density AC/DC PSU for AI servers.
800V capacitor bank units (CBU) with 40W/in³ power density using EDLC super capacitor cells.
An 800V to 12V DC/DC bus converter for compute tray power conversion.
TI will demonstrate the 800 VDC power solution at NVIDIA GTC, held March 16-19, 2026, at NVIDIA's power architecture display and TI's booth 169. The demonstration highlights how TI's analog and embedded processing technologies support advancements in high-voltage systems for AI data centers. Kannan Soundarapandian, vice president and general manager of high-voltage power at TI, stated that the exponential growth of AI computing requires rethinking power delivery in data centers, and the 800 VDC architecture enables operators to address current challenges while preparing for future AI workloads through the collaboration with NVIDIA.
TI's scalable semiconductor solutions extend to areas such as solid-state transformers, sidecar and server IT racks, and cooling distribution units to support future AI infrastructure deployments driven by AI and edge computing growth.
