For years, the conversation about data centers and California’s power grid has followed a predictable script: tech companies want more electricity, utilities struggle to supply it, communities bear the cost, and state regulators try to keep pace with demand that keeps accelerating. Santa Clara is now running a different experiment — one that asks whether data centers could become part of the solution rather than simply the source of the problem.
Silicon Valley Power (SVP), the municipally owned utility of Santa Clara, and Nvidia-backed startup Emerald AI announced a pilot program to demonstrate how flexible data centers can support grid reliability rather than simply drawing from it. The pilot will work with major SVP customers to show that data centers can dynamically adjust power consumption in response to grid conditions without disrupting AI workloads.
If it works, the implications reach well beyond one city’s power lines. California is facing one of the most consequential energy planning challenges in its history — driven almost entirely by the AI industry that Santa Clara sits at the center of. A model that turns data centers into grid assets rather than grid liabilities would change the math considerably.
What the Pilot Actually Does
Under the pilot, Emerald AI will work with major SVP customers to demonstrate data center flexibility using the Emerald AI Conductor platform. SVP will also deploy Emerald AI software to help manage and dispatch participating flexible data centers during limited periods of grid need. The pilot will evaluate how flexible operations can help safeguard reliability, make more efficient use of existing infrastructure, and inform planning for bringing forward phased energization as data center load ramps on the SVP system.
The core concept is demand response at scale. Rather than treating a data center’s electricity consumption as fixed — something the grid must always accommodate regardless of conditions — the Emerald AI Conductor platform allows participating facilities to reduce their power draw during periods of peak load, then ramp back up when grid pressure eases. The platform is designed to make these adjustments without interrupting the AI computations running inside.
The first pilot site will operate at a commercial, multi-megawatt scale at a data center where Nvidia runs AI workloads on advanced GPUs. By utilizing Emerald AI’s software, tightly integrated with Nvidia DSX Flex capability, the data center will precisely respond to SVP signals to support the grid during periods of peak load, while protecting AI workload performance. By proving that advanced computing facilities can dynamically adjust their power usage, the pilot aims to unlock additional capacity across SVP’s network.
That last point — unlocking existing capacity rather than building new infrastructure — is the argument that makes this approach financially and politically viable. California’s grid buildout is expensive, time-consuming, and contested. If demand flexibility can effectively increase the usable capacity of current infrastructure, utilities and ratepayers avoid some of the costs of expansion.
The Voices Behind the Announcement
“Data centers can be more than static electric loads — they can become active partners to the grid,” said Varun Sivaram, founder and CEO of Emerald AI. “We are proud to work with Silicon Valley Power and its customers to demonstrate precise, utility-directed flexibility that data centers can provide in order to better utilize the power system while ensuring they can continue their valuable computing activities.”
Sivaram’s framing — “active partners to the grid” — represents a deliberate repositioning of how data centers are characterized in policy and public discussions. In California, those discussions have grown increasingly tense.
SVP Electric Utility Director Nico Procos said the pilot will help evaluate “practical tools to protect reliability and affordability for our customers while supporting flexible, efficient planning for future load growth.” City Manager Jovan Grogan added that the collaboration “demonstrates how we can turn that momentum into real results for our community.”
Nvidia’s Head of Sustainability Josh Parker said: “As AI scales in Silicon Valley and beyond, the demand for power presents an opportunity for data centers and the grid to become partners.” That framing from Nvidia — one of the companies most responsible for driving the power demand surge — signals that the AI industry is beginning to recognize that its relationship with California’s energy infrastructure cannot remain entirely extractive.
Why California Needs This to Work
Santa Clara is not running this pilot in a vacuum. Across California, the data center buildout tied to AI is straining grid planning at every level.
The state’s Little Hoover Commission issued a 46-page report this year noting that the rapid growth of energy-intensive data centers “presents both a serious challenge and a potential opportunity for California’s electricity system” — coming at a moment when the state already carries some of the nation’s highest electric rates. Communities in Imperial County are gathering signatures for a ballot initiative that would prohibit data centers outright. State Senator Steve Padilla has advanced two Senate bills through committee that would impose new environmental and public health review requirements on large facilities.
The political environment around data centers in California is shifting from cautious acceptance to active resistance in some communities. The Santa Clara pilot represents an attempt to reframe the industry’s relationship with the grid before that resistance hardens into policy that limits where and how fast data centers can be built.
SVP Electric Utility Director Procos testified before the California State Assembly’s joint oversight hearing in January 2026 on the energy impacts of AI and rapid data center growth, sharing how Silicon Valley Power has been planning for increased demand through a $459 million system expansion program designed to double electric load capacity. He also emphasized SVP’s responsibility to protect local ratepayers from the costs of that expansion.
The pilot with Emerald AI is an extension of that planning posture — an attempt to find solutions that serve both the industry and the community it sits within.
A Model Worth Watching
Emerald AI’s work in Santa Clara is its fourth pilot program, following completed efforts in Phoenix, Chicago, and the United Kingdom in partnership with the National Grid. The geographic spread suggests the company is building a playbook applicable across multiple utility environments — not just California.
At CERAWeek 2026, Nvidia and Emerald AI announced a broader collaboration with energy companies including AES, Constellation, Invenergy, NextEra Energy, and Vistra, working to advance what they describe as “flexible AI factories” — data centers designed to connect to the grid faster, generate AI output, and operate as responsive energy assets rather than fixed loads.
California will watch the Santa Clara results closely. If Emerald AI’s Conductor platform can demonstrate that multi-megawatt AI facilities can meaningfully reduce consumption on utility command — without disrupting the computations that make those facilities valuable — it would provide regulators, utilities, and communities with a new framework for thinking about where data centers fit in the state’s energy future.
The alternative is a continued standoff between an industry that needs more power and a grid that cannot keep pace — with California ratepayers and communities caught between them. Santa Clara is betting there is a third path.
Disclaimer: This article is based on publicly available information from the City of Santa Clara, Silicon Valley Power, Emerald AI, and Nvidia, as reported on April 21, 2026. CA Gazette does not have a commercial relationship with any of the companies mentioned.
