COMPUTE,SECURED.
Solomon Data Systems develops AI data center campuses across America: fast-track grid sites built on power that already exists, and behind-the-meter campuses for large scale. Tenants bring the chips. We bring the building, the power and the schedule.
Figures as of October 2026, in critical IT megawatts. Pipeline figures are indicative. Every grid site has signed utility service documents. Project Omega timing is a target.
AI infrastructure will be financed on what it produces, not on what it costs to build. Power quality decides how much useful compute a tenant gets out of every chip it buys. We build for that number.
A flagship campus and a grid portfolio behind it
Large scale where we make our own power, fast track where the power already exists. About 1 GW of critical IT across both.
Our flagship campus: 720 MW of critical IT on its own islanded natural gas plant, with Phase 1 of 360 MW targeted for ready for service in Q2 2028.
- Power
- Reciprocating engines and battery storage · gas supply structured on two interstate pipelines · five days of on-site LNG · grid backup path with access to both the SPP and ERCOT grids
- Status
- Land under term sheet · existing, energized 345 kV switching at the site · 50% data hall design sets complete
Grid campuses built on power that already exists, sized 15 to 100 MW each, to put tenants in production first while large-scale campuses build out.
- Power
- Signed utility documents at every site: electric service agreements, facilities extension agreements or load letters
- Status
- Utility documents signed · site diligence under way · October 2026
Four things, documented
Data center capacity fails for predictable reasons: power that never arrives, land that never permits, equipment that never ships, communities that never consent. Our gate process documents all four before a lease is signed.
Power, documented
Grid service where it already exists, our own generation behind the meter where it does not. Our sites pay for the grid upgrades they require.
Sites, qualified
Land qualified by our own AI siting and diligence platform before we ask for a tenant's time.
Supply, scheduled
Generator sets quoted at six to eight months from purchase order, and switchgear, transformers and cooling at three to nine months by class,¹ bought directly from manufacturers.
Community, in writing
We do not seek local tax abatements, and we offer each host county a binding agreement that carries our commitments on water, light, sound and emissions.
Built for the AI era
Leased by the building, not by the rack. Each 40 MW reference building is engineered for one AI tenant running dense training or inference, and a campus can be one tenant's or several. Not retail colocation retrofitted for the moment.
Designed to 240 kW per rack
Liquid cooled, DC power ready halls designed to 240 kW per rack at steady state, about 85 kW on air and the rest direct to chip, with power delivery sized for AI load steps of up to twice rack power. Built for rack scale systems from more than one chip vendor.
1.1 to 1.2 design PUE ²
No cooling towers and never any evaporative cooling. Less energy overhead and no strain on local water, for the same compute.
Grid grade power engineering
Harmonic management, rack level supercapacitors and blended peaking plus baseload generation. Power quality engineered into the campus, not bolted on.
² Annualized design target. Generation and cooling are sized more conservatively, for the hottest design day at site elevation; final figures depend on site and climate.
A technology company that is also your landlord
Your landlord, never your competitor
We develop the powered, cooled building and stay in it as your landlord. You keep every chip and every customer. We do not run a cloud.
Capacity resale, only on your instruction
Your GPUs run in our halls. If you want to sell spare capacity, we bring buyers on your terms and only on your written instruction. We never operate your GPUs and never share one tenant's information with another. How it works
AI-native from site to lease
Our own AI agents, built in house on commercial foundation models, run land qualification, diligence, underwriting and competitive intelligence.
Three conversations we are ready for
Need megawatts on a real schedule?
Fast-track grid capacity first and behind-the-meter scale next, on timelines the grid alone cannot offer. Discuss capacity
A developer that shows up with proof
We bring our commitments to the first meeting and pay our own way on the grid. Our commitments
GPU capacity and interim space
Resale of tenant GPU capacity at the tenant's written instruction, and interim colocation in partner facilities for tenants who need to deploy before their building energizes. How it works
The power is the product.
Every Solomon campus starts where most projects stall: documented power. Tell us what you need to deploy, and we will show you how fast we can stand it up.
Start the conversation