Before Mark Chung started watching buildings breathe electricity, he was designing the chips inside your phone, your server rack, and your AMD workstation. The pivot was not a rebrand. It was a pool pump.
In 2008, Chung returned from vacation to an electricity bill of $560. His usual tab was around $100. He called the utility. They said the meter was fine. He bought Kill-A-Watt meters, hacked them to run over WiFi, and mapped every circuit in his house until he found it: a broken pool pump drawing four kilowatts - ten times what it should have been.
The fix took minutes. The insight took years to fully land. "The smart way," Chung later said, "would be to just look at one point at all the electricity and unpack it to figure out where it is going." That sentence became a company.
Verdigris Technologies, which Chung co-founded in 2011 with his brother Thomas and their colleague Jonathan Chu, now monitors more than two gigawatts of peak electrical demand across twenty million square feet of global infrastructure - including some of the most critical data centers on Earth. The clients read like a Forbes list: NVIDIA, T-Mobile, Verizon, Google, Qualcomm, AWS, Arm, Westinghouse.
The Pool Pump Problem
A faulty pump drawing 4kW - 10x its rated power - invisible to every dashboard, alert, and utility report. The only way to find it was to look at the electricity itself. Chung built sensors that listen at the breaker box. Everything after that was scale.
The Chips That Made Billions
Spend fifteen years in semiconductor design and you leave fingerprints on a lot of devices. Chung earned his BS and MS in Electrical Engineering from Stanford, graduated in 1999 into Silicon Valley's first great gold rush, and promptly started co-designing chips that would define computing for a generation.
Three different companies. Three significant acquisitions. Each time, Chung stayed the engineer - hands in the architecture, not the deal flow. That changed in 2010, when his son was born.
We've not inherited this planet from our parents, we've borrowed it from our children.Mark Chung — on the principle that drove him to leave semiconductors
The birth of a child reframes everything, but for Chung it arrived with engineering precision. He had spent fifteen years solving problems at the transistor level. Now he wanted to solve a problem at the planetary level - specifically, the $200 billion annual waste embedded in how commercial and industrial buildings consume electricity. Buildings account for roughly 40% of global energy use. Most of that energy leaves no fingerprint. No one knows where it goes.
Chung and Jonathan Chu - who would become Verdigris's CTO and bring machine learning expertise from work in deep-packet inspection - began sketching a system in 2011. The core insight was borrowed from the kitchen table experiment with the pool pump: magnetic sensors clipped at the breaker box, sampling at 8,000 times per second, generating a waveform fingerprint for every device on every circuit. Turn a device on. Train the model. Build a library. Let the AI watch from that single vantage point and read the whole building.
What 8,000 Samples per Second Buys You
The pitch has evolved considerably from the smart-home demo days of 2014. Verdigris now sells electrical intelligence to mission-critical infrastructure - the kind of places where a single UPS failure costs more per hour than some companies earn in a month.
The company's flagship case study involves T-Mobile: across 800-plus UPS rectifiers, Verdigris identified degradation in 4% of units that, in their words, "zero standard alarms had fired on." The projected three-year value from that single deployment: $1.3M to $3M, at a 6:1 ROI. No new hardware. No replacement cycle. Just better data and an algorithm that notices what humans and standard monitoring tools miss.
The 8kHz sampling rate is the technical moat. Most building management systems scrape data every fifteen minutes. Verdigris samples thousands of times per second, capturing sub-cycle electrical behavior - the kind of noise that precedes a transformer failure or flags a cooling unit running at 87% efficiency when it should be running at 100%. It is the difference between an EKG and a clipboard with a once-hourly "how do you feel?" check.
Capacity recovery is the headline metric that enterprise clients obsess over. Verdigris consistently delivers 15-25% stranded capacity recovery in data center deployments. One Fortune 50 NYC data center achieved a 19% chiller energy reduction. In a world where AI infrastructure is constrained by power availability and cooling capacity, those numbers carry strategic weight.
Who Trusts Verdigris With Their Electrons
These are publicly verified clients from Verdigris's own communications and press coverage.
The smart way would be to just look at one point at all the electricity and unpack it to figure out where it is going.Mark Chung — IEEE Spectrum interview, describing the core Verdigris insight
Making the Next Generation's Air Cleaner
Chung does not describe sustainability as a market positioning decision. He describes it as an obligation that crystallized the moment he held his newborn son. Phantom loads - devices drawing power while appearing idle - account for roughly 15% of carbon pollution in buildings. That number is not visible on most energy dashboards. It is invisible by design, or more precisely, by neglect.
His five principles for inspiring younger generations on climate are not the usual corporate talking points. They include building causal understanding - knowing that sea turtle gender is determined by sand temperature is not trivia, it is a map of consequences. And connecting those consequences to things children already love. The strategy is to make the invisible legible, whether in a breaker box or a parent-child conversation about the planet.
He serves on the Board of Directors of ACEEE (American Council for an Energy-Efficient Economy), contributes advice to startups working at the intersection of climate and technology, and has spoken at major infrastructure conferences including the 7x24 Exchange Spring 2025 Data Center Conference on "Future-Proof Data for AI-Driven Data Center Planning and Operations."
In His Own Words
Mark Chung on company structure, OKRs, and building Verdigris:
The Track Record
- Co-developed AMD Opteron - the processor that gave AMD a legitimate claim on enterprise servers
- Contributed to Apple's A7 chip via PA Semi acquisition ($278M exit to Apple)
- Co-developed NetLogic XLP (part of Broadcom's $3.7B acquisition of NetLogic)
- Named Silicon Valley Business Journal "40 under 40" in 2017
- Built Verdigris to monitor 2+ GW of peak demand and 20M+ sq ft globally
- Secured enterprise clients including NVIDIA, T-Mobile, Verizon, Google, Qualcomm, and AWS
- Raised $55M+ in total funding, including a $10M Series B from DCVC in 2023
- Secured strategic investment from Southwire (North America's leading wire manufacturer) in 2026
- Board of Directors member, ACEEE