Google Selects 28 AI Energy Startups for Grid and Data-Center Power Accelerator

Google has selected 28 AI-focused energy startups for its 2026 Google for Startups Accelerator programs in North America and Europe, assembling a cohort aimed at the operational constraints around modern power systems: grid capacity, flexible demand, storage, renewable infrastructure and data-center electricity use. The companies will receive mentorship from Google’s AI and energy experts and access to its digital tools as they work to scale.

The program divides the class between 12 North American startups and 16 European ones. Google frames the effort as part of its work to expand access to clean, reliable and affordable power while supporting growth in its global infrastructure.

Where AI enters the energy system

The startups are not pursuing one common technical approach. Several work at the point where electricity use can be measured, forecast or shifted. Deepgrid, for example, runs an automated virtual power plant intended to coordinate flexible energy sources and help balance the Nordic grid. Elvy manages power flows among a home, the grid and local devices to optimize household energy use.

Other companies focus on finding or using available capacity. Kyra forecasts grid capacity so developers can assess energy projects with less risk, while SMPNet says its real-time network optimization can help utilities increase capacity without new hardware. Optiwatt connects vehicle brands to programs designed to balance energy use and demand across North America.

Data centers are part of the design brief

A distinct group is aimed directly at large new electricity loads. Camus Energy provides software intended to help utilities connect data centers with local energy sources safely and reliably, and says it can shorten that process by three to five years. Felix Fusion helps data centers, mining operations and industrial projects locate energy and connect to the grid more flexibly.

Nyquis similarly targets the time required to power data centers and other large loads. Zendo Energy concentrates on data-center energy usage, seeking to maximize computing performance, while Cosmic Robotics builds autonomous construction equipment for solar farms and data centers. Those products address different stages of the same problem: securing power, building the supporting infrastructure and operating energy-intensive facilities.

A broad portfolio of operating problems

The accelerator also reaches beyond grid software. Capture Energy offers battery-storage software designed to maximize efficiency and revenue. Swish Solar is building an operating system for solar infrastructure, and Reblade uses AI-guided drones to repair eroding wind-turbine blades; the company says its approach is 10 times faster than manual maintenance. Hylight uses hydrogen-powered autonomous airships and AI to inspect power lines and pipelines.

The common mechanism is not simply using AI to generate recommendations. Across the cohort, software and autonomous systems are intended to make energy assets and demand more observable, controllable or maintainable: coordinating flexible resources, predicting capacity, optimizing networks, managing devices, inspecting equipment or automating energy trading. Google’s selection therefore spans both digital control layers and physical energy infrastructure.

Support first, execution next

Over the coming months, Google says founders will receive dedicated technical mentorship from its AI and energy specialists, plus access to Google tools and technologies. The announcement establishes the accelerator’s support model, but the operational outcomes remain with the startups’ individual products: some describe current software capabilities, while others state prospective gains such as faster connections, lower costs or more capacity.

That leaves the meaningful next measure outside the cohort count: whether these tools can move from targeted use cases into sustained utility, home, renewable and data-center operations. Google has aligned the program with its own work on demand response at data centers, energy storage and community energy efficiency, placing the startups in a market where better energy coordination is central to infrastructure growth.