Episode Summary
Executive Summary: William Leyden, co-founder and CEO of Rune, explains how modular, behind-the-meter DC micro data centers can sit directly at solar and wind sites and turn stranded renewable power—especially clipping and curtailment—into compute. The conversation covers Rune’s product design, its Bitcoin beachhead, plans for AI/HPC workloads, and why direct DC-to-DC compute may be faster, cheaper, and more scalable than traditional grid-tied data centers.
Main Topics: Rune’s modular renewable-powered data center design (Priority: 5/5): Rune’s RELIC product is a small, ruggedized, truck-deployable compute cluster built to sit directly at renewable generation sites and consume power without using the grid. Grid inefficiency, clipping, and curtailment (Priority: 5/5): Leyden details how renewable energy is lost today through inverter clipping and grid curtailment, and argues Rune can monetize energy that would otherwise be wasted or shut off. Direct current architecture and bypassing conversion losses (Priority: 5/5): A major thesis is that solar and batteries naturally operate in DC, while conventional data centers force multiple DC-AC-DC conversions; Rune taps high-voltage DC directly for better efficiency and speed. Bitcoin as the beachhead workload (Priority: 4/5): Rune uses Bitcoin mining as its initial use case because it provides an anonymous, permissionless buyer for compute and validates the power-market thesis before moving into broader workloads. Path to AI and interruptible HPC workloads (Priority: 4/5): The company expects to expand into AI inference and other interruptible compute by adding storage and orchestration software, targeting workloads that can tolerate preemption and price-based scheduling. Business model and customer relationship (Priority: 4/5): Rune says its real customer is the power plant: it buys otherwise stranded electricity, deploys its own hardware, and captures value by converting low-value power into higher-value compute outputs. Founder background and company rollout (Priority: 3/5): Leyden’s experience in hydropower, Bitcoin mining, and solar informed Rune’s thesis; the company is already operating on three continents with seed funding and multiple live deployments.
Key Arguments: Renewable generation and data centers were historically designed separately; Rune’s thesis is to co-design compute with energy infrastructure. Solar farms often overbuild DC relative to inverter capacity, creating clipping that can amount to 5%–10% of generation and is usually invisible to asset owners at the DC level. Curtailment creates additional stranded energy when the grid rejects power due to congestion or negative pricing; Rune can consume that power behind the meter. A DC-to-DC architecture avoids unnecessary conversion steps and lets Rune deliver conditioned power to compute faster and cheaper than conventional grid-connected data centers. Batteries are useful but are AC-connected, capital-intensive, and often compete away their own arbitrage spreads; Rune argues compute is a network-effect technology with much larger upside. Bitcoin mining is a practical initial buyer because it is interruptible and always available as an off-take, helping Rune validate the power-side business before pursuing AI/HPC. The longer-term opportunity is to serve interruptible cloud workloads—such as simulations and some inference—via modular compute capacity near renewable plants. Rune’s value proposition is to make renewable plants more profitable without requiring the plant owner to fund major capital expenditures or wait for grid buildout.
Data Points: RELIC power capacity: 100 kilowatts per unit - Each Rune micro data center module is designed around a 100 kW load. RELIC dimensions: Approximately 8 feet long, 2 feet wide, 5 feet high - Physical size of the modular compute unit. RELIC weight: 2,000 pounds fully loaded - Includes compute, cooling, comms, and enclosure. Deployment time: About 45 minutes - Time from truck delivery and forklift placement to energized and cash-flowing operation. Solar facility scale: 500,000 to 750,000 panels - Example of the very large power plants Rune works with. Clipping range: 5% to 10% - Leyden states this is the amount of solar output that can be wasted and monetized by Rune. Power plant example: 300-megawatt solar farm - Used to illustrate how clipping can equal roughly 30 MW of power. Operating duration in Chile: 7 hours per day - Rune data centers in the Atacama Desert reportedly run exclusively on clipped power. Deployment footprint: 3 projects on 3 continents - Rune says it has live projects in Texas, Chile, and Sweden. Company stage: Seed stage - Current financing stage of Rune. Funding/investors: Union Square, Carbon Capital, Vestis - Examples of investors and strategic backers mentioned. High-voltage input range: 800 to 1500 volts DC - Rune says it already taps high-voltage DC directly for compute. AI/compute horizon: Next 18 months - Timeline given for upgrades, onboard storage, and orchestration software to move upmarket. Renewable energy scale reference: More energy hits the earth in one hour of sunshine than humanity consumes in one year - Used to emphasize the scalability of solar as an energy source.
Pivotal Quotes: "We think it has the potential to dramatically accelerate time to compute and also scale compute in a way that maybe you have to go to outer space to do." — William Leyden: Describing Rune’s RELIC modular data center product and its scalability ambition. "There is a hidden power plant behind every single solar power plant, and with Rune, every single solar power plant is a latent data center." — William Leyden: Explaining clipping and Rune’s core value proposition. "Every solar power plant with Rune is a latent data center." — William Leyden: Summarizing the company’s thesis that renewable sites can become compute infrastructure.
Implications: Rune points to a future where compute is colocated with energy production, reducing conversion losses and grid dependence. If successful, renewable plants could become distributed AI/compute infrastructure, changing power market economics and speeding data center deployment.