Executive Summary
GridCARE, a startup using AI to unlock unused grid capacity for data centers, raised $64 million in a Series A round, led by Sutter Hill Ventures (an early Nvidia investor) with backing from venture capitalist John Doerr, bringing total 77.5 million. The company’s software analyzes power grids across millions of scenarios to identify underused capacity, allowing data centers to connect faster than waiting for new generation or transmission—potentially in years rather than decades.
The Players
GridCARE was founded by CEO Amit Narayan and has backing from Ram Shriram, an early Google investor who is also a GridCARE board member and seed investor. The company is not building power plants, buying interconnection rights, or converting crypto mines—it is selling software and analytics to utilities and hyperscalers.
Sutter Hill Ventures is the lead investor, known for its early, highly profitable bet on Nvidia. John Doerr (via Kleiner Perkins, though the article does not specify his vehicle) is a legendary Silicon Valley VC known for green tech and internet bets. Their involvement signals that grid software, not just hardware, is now a tier-one AI infrastructure opportunity.
Portland General Electric (PGE) is the first disclosed utility partner, working with GridCARE on a plan to unlock up to 400 MW of excess capacity in Hillsboro, Oregon, by 2029, serving six data centers.
So What?
This deal validates two critical trends:
First, the constraint is no longer just generation—it’s grid intelligence. GridCARE is saying you don’t need new steel. You need better software to use what already exists. Stanford research shows only one third of the grid is used most of the time. The other two thirds is stranded—for now.
Second, peak demand, not average demand, is the real bottleneck. Utilities cannot connect a 100 MW data center if it would cause blackouts during the 10 hottest hours of the year. GridCARE’s AI models determine how many hours a data center can be interrupted (curtailed) to protect the grid. This is not firm power—it’s interruptible capacity. But for some AI workloads (training, batch inference), interruption is acceptable if the price is right.
What Should You Do With This?
If you are a utility: GridCARE represents both a threat and an opportunity. The threat: hyperscalers may bypass you entirely with behind-the-meter generation. The opportunity: you can use software to monetize your stranded capacity without building a single new wire. Every utility should be evaluating grid-enhancing technologies (GETs) and dynamic line rating software immediately.
If you are a data center operator with a power shortage: do not assume you need to buy a power plant. First, hire a firm like GridCARE (or build internal capability) to analyze your target grid’s actual usable capacity at 95% or 98% reliability, not nameplate. You may find 200 MW where you thought there was zero.
If you are an investor in power assets: the GridCARE funding round is a warning. Software-defined grid optimization could reduce the scarcity value of the assets you are buying. If AI unlocks 20-30% more capacity from existing wires and plants, the premium for new-build generation may compress. Hedge your hardware bets with software exposure.
If you are a policymaker: current interconnection queues are broken. The problem is not just physical capacity. Fund and mandate dynamic line rating, advanced power flow control, and AI-driven curtailment analysis. You can triple effective grid capacity faster than you can build one new transmission line.
Source: Bloomberg, May 14, 2026
