Artificial-intelligence infrastructure is running into an increasingly unusual constraint: Electricity can exist without being available where computing companies need it. Rune, a California startup, is betting that some of that mismatch can be solved by moving the computers instead of waiting for the power network to catch up. The company has unveiled RELIC, a modular compute platform designed to sit directly alongside operating solar facilities and consume electricity before the generation reaches the conventional grid. The launch comes with a $40 million Series A led by Spark Capital, taking Rune’s total funding to $53.5 million. Rather than approaching the data-center market through another large construction project, Rune is targeting the gap between renewable generation and the infrastructure capable of monetizing every electron. That strategy places the company at an intersection between solar economics, AI capacity and the increasingly valuable question of how quickly computing equipment can reach usable power.

Rune Targets Stranded Solar Electricity

The basic premise behind RELIC is that renewable generation does not always translate into electricity that the grid can immediately absorb or economically reward. Rune says solar facilities can waste as much as 20% of their generated electricity, representing more than 50 terawatt-hours of annual potential in the US. Instead of sending that unused electricity through the conventional grid pathway, RELIC connects computing capacity directly to the solar installation. That arrangement allows Rune to avoid some of the metering, utility and interconnection constraints that have become major considerations for new AI infrastructure projects. The company is therefore treating stranded generation as a computing resource rather than simply an inefficiency within the power system. The distinction matters because AI developers can face lengthy waits for new grid connections even when generation already operates nearby.

“Every solar plant is a latent data center. The power is already there, sitting idle while AI labs wait years for grid connections that may never come,” said William Layden, Co-Founder and CEO of Rune. Layden’s argument reframes the infrastructure problem around access rather than simply total electricity supply. A solar project can generate power, yet that generation may still fail to provide a practical route for a new computing load because transmission capacity, interconnection queues and conventional construction schedules impose additional constraints. RELIC attempts to remove several of those steps by positioning the computing hardware at the source of generation. The model also changes the commercial relationship between renewable operators and computing buyers, potentially giving solar assets another way to monetize output that otherwise has limited value. For Rune, that possibility forms the foundation of a market that could expand alongside both solar generation and AI workloads.

RELIC Brings Compute Behind The Meter

Rune designed RELIC as a modular system rather than a conventional data-center building, allowing the company to place computing equipment on existing renewable infrastructure with minimal physical intervention. The company says installation can take about 60 minutes, while customers can receive working GPU capacity within six weeks of signing a contract. That schedule compresses several activities that normally stretch across much longer development cycles, including land preparation, power infrastructure, grid connection and facility construction. Rune says its Texas installation demonstrates the approach at an operating 200-megawatt solar facility, where the system uses existing infrastructure without modifications to the underlying site. The company says the installation requires no extensive site preparation and leaves no visible data-center footprint comparable with a traditional facility. In practical terms, Rune is attempting to make compute deployment resemble equipment installation rather than a major real-estate development.

The system also takes a different approach to electricity conversion by operating natively on direct current. Solar panels already produce direct current, while conventional electrical infrastructure typically converts that output into alternating current before delivering it through the grid. RELIC instead seeks to keep computing closer to the original electrical form, reducing conversion stages and the associated equipment requirements. Rune says the architecture can lower non-compute infrastructure spending by 85%, with estimated savings of roughly $620 million across a 100-megawatt deployment. Those figures remain company claims rather than independently verified industry benchmarks, but they highlight the economic proposition behind the system. If direct access to solar electricity can reduce both equipment requirements and development time, the economics of small renewable-sited computing facilities could look materially different from those of conventional AI campuses.

A 200 MW Texas Facility Tests The Model

Rune’s initial working deployment gives the company a physical reference point for its modular strategy. The RELIC installation operates at a 200-megawatt solar facility in Texas, using existing infrastructure and avoiding modifications to the solar site, according to the company. Texas provides a significant environment for testing the model because the state has developed substantial renewable generation while also supporting a rapidly expanding market for data-center and AI infrastructure. The combination creates a setting for testing whether computing loads can absorb electricity that might otherwise have limited commercial value. Rune’s deployment provides a live example of how computing equipment can operate directly at an existing renewable asset without new grid work or construction. The approach could therefore give solar operators another potential way to monetize underutilized generation without requiring them to transform their facilities into conventional data-center campuses.

The company’s strategy also responds to a broader mismatch between the physical pace of data-center construction and the speed of AI demand. Large facilities require land, electrical equipment, cooling systems, construction labor and utility coordination, with each layer capable of creating a separate schedule constraint. RELIC is designed to reduce several of those requirements by using existing renewable generation as its power source. Rune says the system does not consume water, while its compact architecture avoids the extensive construction associated with a conventional facility. The approach could therefore allow computing deployments in locations where a full-scale data center would require substantially greater infrastructure. Yet the model still depends on suitable solar assets, reliable hardware operations and customers willing to place workloads within a nontraditional computing environment.

Rune Bets On Direct-Current Computing

The direct-current architecture sits at the heart of Rune’s technical proposition because it links the electrical characteristics of solar generation with those of modern computing hardware. Solar panels already produce direct current, while conventional electrical infrastructure typically converts that output into alternating current before delivering it through the grid. RELIC instead seeks to keep computing closer to the original electrical form, reducing conversion stages and the associated equipment requirements. Rune says the architecture can lower non-compute infrastructure spending by 85%, with estimated savings of roughly $620 million across a 100-megawatt deployment. Those figures remain company claims rather than independently verified industry benchmarks, but they highlight the economic proposition behind the system. If direct access to solar electricity can reduce both equipment requirements and development time, the economics of small renewable-sited computing facilities could look materially different from those of conventional AI campuses.

Varun Palivela, Rune’s Co-Founder and CTO, described the broader design philosophy this way: “AI’s power constraint isn’t generation, it’s resource allocation,” said Varun Palivela. “The industry is retrofitting data centers designed for a different era rather than rethinking the architecture itself. We built an entirely new technology stack from the ground up, coupling compute directly to clean generation so AI can scale here on Earth.” His comments point to the architectural problem Rune believes conventional facilities have inherited from earlier computing cycles. AI systems impose different power-density, deployment-speed and hardware requirements than many facilities originally designed for general-purpose computing. Rune is attempting to address those requirements by designing the power and compute relationship as one system rather than as separate infrastructure layers.

Investors See A New Compute Supply Route

Spark Capital led Rune’s $40 million Series A, with the financing bringing total capital raised by the company to $53.5 million. The investment gives Rune additional capital as it expands its RELIC platform and builds relationships with renewable-energy partners and compute customers. Santo Politi, Founder and General Partner of Spark Capital, framed the opportunity around the widening gap between AI infrastructure demand and the physical systems needed to satisfy it. His assessment also suggests that investors see renewable assets as potential locations for distributed compute rather than merely sources of electricity for distant data centers. A distributed network of solar-sited compute could potentially grow through repeated installations rather than through a small number of large campuses. That model would shift some infrastructure spending away from buildings and toward standardized modules, power electronics and computing hardware.

“Demand for AI infrastructure is growing faster than it can be delivered. Rune deploys compute directly behind the meter at operating solar farms, converting clean power into rapidly scalable AI capacity,” said Santo Politi, “Every renewable asset in operation becomes a potential deployment site. Rune is the fastest path to new capacity we have seen.” Politi’s statement captures the investment case around deployment speed, although the broader market will determine how widely that advantage translates into commercial adoption. A distributed network of solar-sited compute could potentially grow through repeated installations rather than through a small number of multibillion-dollar campuses. That model would shift some capital expenditure away from buildings and toward standardized modules, power electronics and computing hardware. In turn, the economics could become increasingly dependent on manufacturing scale and the ability to match individual renewable sites with suitable AI workloads.

Solar GPU Datacenters Could Redefine Build Cycles

Rune’s proposition ultimately depends on whether AI infrastructure buyers will accept a different definition of a data center. RELIC does not attempt to recreate the large, centralized facility in a smaller box; it removes several elements of that model and places computation directly beside generation. The company says that strategy can produce GPU capacity within six weeks of a contract while installing the physical system in roughly an hour. Conventional projects often require years of planning and construction, making the potential difference significant for companies racing to deploy AI services. The model also offers a way to connect the growth of renewable generation with the growth of computing demand without automatically requiring a new generation project for every new computing cluster. Rune’s approach could create a parallel category of modular, renewable-sited computing if its deployment claims translate into repeatable commercial projects.

Still, the concept faces a practical test that extends beyond installation speed and capital efficiency. Solar generation changes throughout the day, while AI customers increasingly require predictable capacity, especially for applications that depend on consistent latency and availability. Rune therefore needs to demonstrate that its system and customer model can make variable renewable power useful at commercial scale. The company’s initial Texas deployment provides an important starting point, but a broader market will require repeatable installations across different solar configurations and operating conditions. Success would not mean traditional data centers disappear, because large AI clusters will continue to require substantial power, cooling and network infrastructure. Instead, Rune is testing whether a parallel category of modular, renewable-sited computing can absorb electricity that the existing system struggles to use and convert it into AI capacity on a faster deployment cycle.