Energy-abundant corridors
West Texas and the Permian, Appalachian gas, the Upper Midwest. Slower fiber, faster electrons.
Infrastructure, compute, energy.
Capital at the table.
Obsidian is an origination and development firm operating across the four markets behind AI infrastructure — the site, the power that serves it, the silicon that fills it, and the capital that funds all three. One counterparty across all four layers. We are not a marketplace, and we do not publish a price list.
The land people don't speak turbines. The GPU people don't read interconnection queues. The capital asks a question nobody on the deal can answer, and ninety days evaporate.
It is a site with a defensible interconnect position, generation deliverable on the schedule the tenant underwrote, hardware that is contracted rather than promised, and capital that understands all three well enough to price the risk. Obsidian works across all four.
A signed LOI from a utility is not an energization date. The gap between them has killed more campuses than capital ever has.
Allocation gets sold against capacity that will not be energized in time to take delivery of it.
Lenders underwrite depreciating hardware against a tenant covenant nobody has read closely.
Four advisors, four fee structures, and no single party accountable for the transaction actually funding.
A marketplace is paid for introductions and is indifferent to what happens after them. Obsidian is paid for closings. We diligence before we introduce, we work one side of a deal at a time and we say which side we are on. Where a transaction warrants it, we take a development or ownership position rather than a fee.
We work the capital question first, not last. A transaction that cannot be financed isn't a transaction — it's an introduction with a deadline on it.
Obsidian holds live positions across each layer. A mandate can start at any one of them — most start at power.
Powered land, powered shells and operating facilities, underwritten on energization date, queue position and deliverability rather than a brochure.
Queue position and study status. Interconnection agreement, executed or not. Title, easements and water rights. Electrical topology to the rack. We would rather kill a site in week two than introduce it in week ten.
Allocation is only worth what the ship date and the paper behind it are worth. We verify both before anything reaches a buyer.
Chain of title on the allocation. Named ship windows rather than quarters. Whether the OEM relationship survives assignment. An allocation that cannot be energized on schedule is inventory, not capacity.
Power is the binding constraint on every transaction in this market. It is the first thing we diligence and the last thing we take on faith.
Firm versus interruptible, and what the contract says under load. Fuel supply and transport. Equipment lead times against the construction schedule. Every schedule in this business is a power schedule wearing a different hat.
Capital is not a referral we make at the end. It sits at the table from the first conversation, which is why our deals are structured to be financeable before they are marketed.
Tenant covenant, contract tenor, residual assumptions, and where in the stack the risk is genuinely sitting. A deal that cannot be financed is not a deal; it is a press release waiting to be retracted.
Firm or interruptible, fuel and transport, equipment lead times against the construction programme. Everything else in the transaction moves when this does.
What is being sold, bought, built or financed — and the constraint that governs it. Scope, side and fee in writing before we start.
Power, title, queue, paper, counterparty. We tell you what we found whether or not it helps the deal.
The shape of the transaction and where capital sits in it. Financeable before it is marketed to anyone.
Negotiation, documentation and hand-off into construction or operations. We stay on the deal until it funds.
Where the opportunity warrants it, we take a development or ownership position rather than a fee.
The end of a transaction we worked is a building that energizes on the date the tenant underwrote.
West Texas and the Permian, Appalachian gas, the Upper Midwest. Slower fiber, faster electrons.
Power already built for something else. Legacy industrial load, flare gas, and interconnects worth more than their hashrate.
Northern Virginia, Dallas, Phoenix, Chicago, London. Power is scarce, which is why a defensible queue position is worth underwriting carefully.
Sites, shells and campuses looking for tenants, capital or an exit.
Buyers of capacity who need it energized on a date they can plan against.
Powered assets repositioning from hashrate or heavy industry into AI workloads.
Generation, storage and equipment positions looking for an offtaker with a schedule.
Capital seeking diligenced exposure to the buildout rather than a pitch deck.
Supply-side partners with inventory, allocation or deployment capability.
Queue position, study status, title, topology. The work that decides whether a transaction is real happens before anyone is introduced.
A derelict plant with a live interconnect beats a clean parcel with none. Knowing which is which is most of the work.
Nothing reaches a counterparty until we have read the paper ourselves. If diligence weakens the deal, you hear it from us first.
We represent the buyer or the seller on any given transaction, never both, and the mandate states which before work begins.
Assets, allocations and capital sources are not shopped. Nothing moves, and no name is used, without written permission.
If an asset doesn't clear our own underwriting, we decline the mandate rather than pass the risk to someone who trusts us.
Land is available. Capital is available. Silicon is available. What is scarce is the party who can clear all four constraints at once and put a date on it. That is the whole of our business.
Tell us what is actually blocking the transaction — the power, the paper, the allocation or the money. We will tell you within a week whether we can move it, and we will say no when the answer is no.