#United States #Dallas #NVIDIA #Texas Instruments #800VDC

Introduction

Texas Instruments (TI), a leader in semiconductor design and manufacturing, has made significant strides in enhancing the power architecture for AI data centers. At the NVIDIA GTC event, which runs from March 16 to 19, 2026, TI unveiled its latest innovation: a complete 800V direct current (DC) power architecture. This breakthrough is positioned as an essential component for the next generation of AI infrastructure, demonstrating how TI’s technology synergizes with NVIDIA’s high-performance processing capabilities.

A Shift in Power Delivery

As the demand for AI-powered applications surges, the current infrastructure for power delivery is being tested like never before. Kannan Soundarapandian, vice president and general manager at TI, highlights the need for a radical rethink of how power is supplied to AI computing systems. The company's 800VDC architecture aims to tackle the pressing power challenges faced by data centers today, enabling them to efficiently meet the rising power requirements of AI workloads.

Efficient Architecture

The innovative architecture designed by TI introduces several key advantages:
1. Maximized Conversion Efficiency: By employing a streamlined design that requires only two conversion stages, TI’s solution converts 800V directly to processor power. This significantly enhances efficiency across the entire power path, reducing the energy lost during conversion.
2. High Power Density: The architecture allows for compact designs that maintain high power density, which is crucial for the demanding environments of AI data centers.
3. Scalability: This architecture is engineered to scale seamlessly, enabling data centers to adapt to increasing workloads without compromising performance or reliability.

Features of TI’s 800VDC Power Architecture

Some highlights include:

• - 800V Hot-Swap Controller: To ensure robust input power protection for the 800V rails, allowing for continuity even during maintenance or component changes.
• - 800V to 6V DC/DC Bus Converter: This high-density solution uses integrated GaN power stages, achieving an impressive 97.6% peak efficiency, which is ideal for compute tray applications.
• - 6V to <1V Multiphase Buck Converter: Tailored for advanced GPU cores, this high-current solution improves upon traditional 12V designs, offering higher power density and superior performance in demanding conditions.

Additionally, TI will display a 30kW 800V high-power density AC/DC power supply unit (PSU) designed specifically for AI servers alongside innovative capacitor bank units featuring eco-friendly EDLC super capacitor cells.

Partnership with NVIDIA

TI’s collaboration with NVIDIA is a natural alignment of two industry giants looking to push the boundaries of what’s possible in AI infrastructure. The goal is to accelerate AI workloads with a robust foundation built on advanced semiconductor technologies. TI’s power solutions will not only elevate the efficiency of AI data centers but also encourage the development of new, scalable applications, laying the groundwork for future technological advancements.

Conclusion

The unveiling of TI’s comprehensive 800VDC power solution at NVIDIA GTC marks a significant milestone in the evolution of data center infrastructure designed to meet the needs of future AI systems. By prioritizing efficiency, scalability, and innovation, Texas Instruments is setting the stage for a new era in power delivery—one that can accommodate the explosive growth of AI while maintaining reliable operations. As these advancements materialize, they will undoubtedly influence data center designs and operations worldwide, paving the way for smarter, more efficient computing solutions that can handle the requirements of tomorrow's technology.

For more details on TI’s product offerings and innovations, visit ti.com.