TI and the Future of AI Power: Smarter Systems from Grid to Gate

Dec 3, 2025

Chris Suchoski, who leads data center systems engineering at Texas Instruments, shares insights on the evolving landscape of AI power solutions. He discusses the urgent need for faster design cycles in the face of burgeoning AI workloads, detailing how traditional power architectures must adapt. The shift towards high-voltage systems like 800V is explored, emphasizing its implications for cooling and infrastructure. Chris also highlights TI's unique position in leveraging automotive experience to meet new challenges and the significance of collaboration through OCP.

Ask episode

AI Snips

Chapters

Transcript

Episode notes

INSIGHT

Compressed Timelines With Rising Demands

AI demand has compressed product development cycles from years to under 12 months.
Power, density, sustainability, and reliability requirements still increase despite the shorter timelines.

INSIGHT

800V Shift Affects Whole Facility

Power architecture shifts to 800V (±400V) are pushing changes beyond the compute tray to the whole data center.
This affects distribution, cooling, storage, and backup systems across the facility.

INSIGHT

Unpredictable AI Loads Stress Power

AI workloads are aperiodic and unpredictable compared with traditional compute.
That unpredictability stresses power delivery and the wider grid in new ways.

Get the Snipd Podcast app to discover more snips from this episode

Get the app

Texas Instruments is known for semiconductors, calculators, and decades of design expertise. In this episode, Chris Suchoski shows how TI is becoming a full grid-to-chip partner for the next wave of data center growth. Rob and Chris break down what it takes to power AI, how architectures are shifting, and why TI’s deep experience outside the four walls of the data center matters now more than ever.Chris outlines the pressure facing the industry. AI workloads are growing at a rate the traditional design cycle cannot match. Time-to-market has gone from years to months, yet the expectations for density, power, sustainability, and reliability continue to climb. TI’s response is to support the entire power path from the grid to the GPU. Their goal is to help customers move faster and build systems that can take on the new wave of compute demand.They discuss the shift from conventional rack design to high-voltage architectures like 800V, the rise of megawatt-class racks, and what happens when unpredictable AI workloads collide with an aging set of power assumptions. Chris explains why this shift impacts every part of the ecosystem, from AC distribution to cooling to storage systems. He also talks through how TI’s decades of work in automotive, industrial, and grid power translate directly into the needs of hyperscale and AI operators.The conversation also highlights how collaboration through OCP speeds progress. Standards development, shared learning, and early visibility into architectural trends help companies build aligned solutions. Chris explains why TI participates heavily in OCP working groups and why he sees it as one of the best ways for operators and engineers to keep up with rapid change.Chapters:00:00 Introduction01:00 Meet Chris Suchoski and TI’s data center focus03:00 Why timelines are shrinking04:00 AI workloads and unpredictable power demand06:00 TI’s role beyond the compute tray08:00 Grid-to-chip: TI’s end-to-end approach10:00 Lessons from automotive and grid power12:00 Why OCP collaboration matters14:00 The shift to 800V architectures17:00 What Gen4 power architectures may look like19:00 Cooling, reliability, and emerging risks21:00 TI’s latest hardware and reference designs24:00 Hidden bottlenecks and future challenges26:00 What operators and engineers should prepare for28:00 How to engage with TI and get involved in OCP