At least one U.S. data center builder has begun bypassing the electrical grid entirely, turning to modular solar microgrids to get facilities online faster. The move reflects a growing tension between surging demand for AI computing power and the years-long waits required to connect new projects to the nation’s transmission infrastructure. A wave of startup funding and strategic partnerships now aims to turn that workaround into a scalable industry.
Grid Queues Are Choking Data Center Growth
The core problem is straightforward: the U.S. power grid cannot process new connections fast enough to keep up with demand. The latest interconnection data from Lawrence Berkeley National Laboratory, summarized in its 2025 “Queued Up” report, documents the scale of the bottleneck. The dataset and analysis track U.S. interconnection queue volumes, timelines, and the structural delays that slow new generation and large-load projects from reaching the grid.
Those delays act as a binding constraint on data center construction. A facility that needs hundreds of megawatts of power cannot simply flip a switch; it must wait in line behind thousands of other proposed projects, many of which will never be built but still clog the queue. For hyperscale operators racing to deploy AI training clusters, every quarter of delay translates into lost revenue and competitive disadvantage. The mismatch between the speed of chip innovation and the pace of grid permitting has created a structural gap that traditional utility planning was never designed to close.
Energy market analysts have begun treating that gap as a defining feature of the AI buildout. Teams at Bloomberg’s energy research group, for example, have highlighted how grid constraints are reshaping where and how large computing loads are sited. Their work underscores that the problem is not just about adding more generation; it is about the slow, capital-intensive process of expanding transmission and distribution infrastructure to deliver that power where it is needed.
Exowatt’s $70 Million Bet on Off-Grid Power
Into that gap steps Exowatt, which closed $70 million in Series A funding to deploy and scale its Exowatt P3 platform. The system uses portable solar panels mounted on single-axis trackers, designed for rapid setup at sites that lack grid access. Rather than waiting years for a utility interconnection agreement, a data center operator can begin generating power on-site within weeks of breaking ground, at least for a portion of its total load.
The capital raise signals that investors see off-grid solar for data centers not as a niche experiment but as a viable path to faster returns. Exowatt pitches the P3 hardware as a modular asset that can be deployed in phases, letting customers match capacity additions to their own construction timelines. The funding will help the company standardize its hardware, streamline logistics, and build out the field teams needed to install and maintain arrays at multiple sites simultaneously.
According to Exowatt’s disclosures, the new money is earmarked not only for manufacturing but also for software and controls that optimize how each array operates. The company’s DG Matrix and Exowatt, announced in early February 2026, to deliver gigawatt-scale dispatchable power for the AI era. DG Matrix brings power electronics, distribution technology, and control software. Exowatt contributes its modular solar generation platform. Together, the companies aim to aggregate many individual microgrid installations into a coordinated power supply large enough to serve clusters of data centers.
The word “dispatchable” matters here. Solar generation alone is intermittent, producing power only when the sun shines. Data centers, by contrast, run around the clock and penalize even brief outages. Any serious off-grid strategy must pair solar panels with energy storage or backup generation that can deliver electricity on demand, day and night. The DG Matrix partnership suggests the companies are building toward that full-stack solution, though publicly available details on the specific storage or backup technologies involved remain limited.
Scaling to gigawatt levels also raises operational questions. Coordinating dozens or hundreds of modular plants requires sophisticated monitoring, forecasting, and maintenance workflows. Enterprise buyers will expect robust service frameworks, including clear channels for
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
