top of page
  • X - Sean Walsh
  • LinkedIn

Chapter 10: Solution - Solar Computing Clusters

Updated: 1 day ago

My patented Solar Computing Clusters can solve the American electricity shortage. What are they? How do they work?


639 Solar Computing Node - First licensee of one of my patents
639 Solar Computing Node - First licensee of one of my patents


"Simplicity is the ultimate sophistication." - Steve Jobs



Key Points:


1. Solar Computing Clusters appear to be the only system capable of solving America's enormous (44GW+) electricity shortage by 2028.
2. The first company to license one of my patents is one called 639 Solar, that I cofounded. We successfully launched our first commercial scale off-grid solar/battery datacenter in New Mexico in 2022.
3. With continuous uptime and incredibly low electricity costs, 639 Solar has proven the commercial success of Solar Computing Clusters, which many pundits dismissed as "impossible".

4. My Solar Computing Cluster intellectual property extends far beyond that which has been commercialized by 639 Solar, and great opportunity awaits any company that shares my vision and is able to invest the resources for large scale deployment.
5. Solar Computing Clusters aren't just faster time-to-power. Surprisingly, they also yield a lower electricity cost than most, perhaps all, alternatives.


Key Stats:


Below $0.03/kWh - 639 Solar's electricity production cost.

24hrs x 365days - datacenter uptime inherent in Solar Computing Cluster system, and proven by 639 Solar.

5x - 639 Solar's approximate ratio between total solar array production, and baseload datacenter capacity.

22MWhr - approximate battery storage capacity per 1MW of baseload datacenter capacity, in current 639 Solar system.



Don't Take My Word For It


In December 2024, a team of expert researchers from Scale Microgrids, Stripe, and Paces published a white paper that sent shockwaves through the datacenter industry. Their question was simple: Could off-grid solar micro grids be big enough, fast enough, and cheap enough to power the AI race, thus solving America's electricity shortage?


Their answer was unequivocal: Yes.


Their findings were exciting. An off-grid solar microgrid delivering 44% of energy from renewables would cost approximately $93/MWh—just $7 more than off-grid gas turbines at $86/MWh. A system delivering 90% renewable energy would cost $109/MWh—significantly cheaper than Microsoft's contract to restart Three Mile Island nuclear at $130/MWh.


Projections from OffGridAI.us
Projections from OffGridAI.us

The OffGridAI's modeled results easily justify the merit of off-grid solar/battery computing.


More importantly, OffGridAI identified the decisive time-to-power advantage offered by off-grid solar/battery computing. While grid interconnection takes 5+ years and gas turbine procurement requires as much as 7 years, they predicted that an off-grid solar micro grid can reach operation in approximately 2 years—and motivated builders could compress that further.


The researchers identified more than 1,200 GW of suitable datacenter capacity in the American Southwest alone—enough to cover 4-40x all projected U.S. datacenter growth through 2030.


What they didn't mention is that someone had already designed such a system. Someone had secured the patents. And, someone had already commercialized a system similar to what they advocated.


And the most surprising part... the actual commercial results were even better than what the OffGridAI experts modeled. The OffGridAI team projected that solar/battery could reasonably provide up to 90% of the total datacenter electricity. However, 639 Solar proved it could effectively all the way to 100% solar/battery, while still achieving an LCOE below that of all known legacy power sources in the market.


The OffgridAI felt that 90% electricity from solar, at a cost around $110/MWhr was more than good enough to justify large scale focus. But, the 639 Solar off-grid datacenters achieved 100% solar, at an over 80% discount from the OffGridAI models. 639 Solar achieved a cost of just $30/MWhr.



Trinity Redux - Another Success In The Southern New Mexico Desert



On July 16, 1945, at 5:29 a.m. Mountain War Time, the atomic age began in the Jornada del Muerto desert of southern New Mexico. The Trinity test—the first detonation of a nuclear weapon—released 18.6 kilotons of power, vaporizing a 100-foot steel tower and turning the surrounding sand into green glass.


Seventy-seven years later, just a few miles west of that same desert, another kind of breakthrough occurred, albeit more quietly.



In December 2022, 639 Solar's first Solar Computing Cluster—a commercial-scale prototype embodying a new approach to datacenter power—was energized. No fanfare. No press releases. Just electrons flowing from photovoltaic arrays through battery storage into computing equipment, continuously, around the clock, independent of any external grid.


The system works.


It has been operational for years. And the intellectual property protecting it has been methodically secured through seven issued U.S. patents covering every critical aspect of the technology.




Inside The 639 Solar Computing System



Traditional datacenter power flows through a labyrinth of inefficiency: generation plant to high-voltage transmission lines to substations to distribution networks to facility transformers to computing equipment. At each step, electricity is lost.

639 Solar's system architecture bypasses this entire chain.


639 Solar's modular system begins with a solar PV array sized at roughly five times the baseload datacenter capacity. Battery storage of approximately 22 MWh per MW of baseload datacenter provides the buffer that enables 24/365 operation.



We made a number of critical engineering choices with the 639 Solar system, which combine to yield its outstanding performance:


  1. The system is designed around battery technology that costs 80% less than lithium-ion, and is more easily recyclable.

  2. Low solar array ground coverage ratio provides compound benefits.

  3. Maximize DC power bus to minimize electricity losses.


The result: production costs below $0.03/kWh—among the lowest electricity prices in the United States. No fuel costs. No grid charges. No capacity payments. And because maintenance costs are a tiny fraction of fuel costs for traditional power plants, these electricity costs are effectively fixed for over 30 years.




The Patent Portfolio: Comprehensive Protection


The intellectual property securing Solar Computing Clusters is not a single patent covering a narrow invention. It is a comprehensive portfolio of issued U.S. patents protecting the entire design space of solar-powered high-performance computing.


The coverage spans five distinct layers:


  1. Foundational System Architecture — The physical and logical topology for directly powering high-duty compute clusters from renewables plus storage

  2. Power Distribution, Management, and Optimization — The control layer that decides how much power to route from each source to each load

  3. Energy Storage Design and Operation — How the storage system is integrated, monitored, and controlled

  4. System Sizing and Design Envelope — Quantitative ranges for solar, storage, and compute capacity ratios

  5. Commercial Applications — Explicit coverage of AI/ML, cloud, HPC, and cryptocurrency mining


Patent Coverage Matrix:

Solar Computing Cluster Patents - Sean M. Walsh
Solar Computing Cluster Patents - Sean M. Walsh


Window of Opportunity - Escalating Value - The Perils of Inaction


The opportunity described in this chapter is not likely to remain available for long. One or more market participants will awaken and move to secure market dominance leaving others destined for consolidation or worse.


The 44 GW shortfall exists because demand is growing faster than supply can respond. Every month that passes without new capacity increases the scarcity premium. Any hyperscaler that secures incremental power gains a competitive position that late movers may never overcome.


The patent protection securing Solar Computing Clusters will become more valuable as the industry recognizes what independent researchers have already concluded: off-grid solar microgrids are "likely the only clean solution that could also achieve the scale and speed requirements" for AI infrastructure.


Operators who attempt to build solar-powered datacenters without addressing the intellectual property landscape will find themselves navigating claims that cover the architecture they need, the control systems they require, the storage integration they depend on, and the sizing parameters their economics demand.


I invite strategic acquirers to evaluate my patent portfolio's fit with their broader infrastructure ambitions.


The path forward is simple: solve America's electricity shortage, collect enormous profit, and ensure American AI supremacy.


The technology that can power America's AI future is not theoretical. It's been moving electrons for years in the New Mexico desert.



For more information on Solar Computing Clusters, the patent portfolio, and partnership opportunities, contact me on LinkedIn.






Sources:

  1. Fast, Scalable, Clean, And Cheap Enough: How Off-Grid Solar Microgrids Can Power The AI Race, OffGridAI

  2. Off-Grid Microgrids: The Future Of Sustainable Datacenters, Datacenter Magazine

  3. US Patent 10,795,428B2, Sean Walsh

  4. US Patent 11,289,914B2, Sean Walsh

  5. US Patent 11,929,622B2, Sean Walsh

  6. US Patent 11,962,157B2, Sean Walsh

  7. US Patent 11,967,826B2, Sean Walsh

  8. US Patent 12,081,030B2, Sean Walsh

  9. US Patent 12,126,179B2, Sean Walsh

  10. www.639solar.com




bottom of page