Independent advisory · AI infrastructure
We do not buy or sell anything. We provide access.
Access to every layer of AI infrastructure, at scale — data center capacity, GPU allocation, GPUaaS agreements, power, and the offtakers on the other side of the table. We work for the organizations vendors will take a meeting with but will not give the terms they give the giants. We hold no inventory and take no position in any deal, and where a deal needs more than a single piece we can bring all of it.
We act for the party running the workload.
Stated up front so nobody spends a week finding out. The restriction is not posturing — it is the reason operators quote us real numbers instead of screening numbers.
A file opens for the party whose workload occupies the capacity and whose balance sheet stands behind the contract, or an advisor that party names in writing. Not brokers, not resellers, not intermediaries assembling a position to flip.
You contract directly with the operator, developer or channel. We are not in the chain and we do not add one. An allocation already shopped through four parties is one we pass on — and so will the operator, which is the part that actually costs you.
The availability board reads without an email address and every mandate page says plainly what the work is. A conversation gets you a number against your specific requirement — which is a different thing from a price list.
GPUaaS and reserved clusters are current-generation Blackwell only, from 512 GPUs or 1 MW upward, no ceiling; hardware allocation runs wider. Below that floor we would be adding cost without adding much, and we would rather say so than take the meeting.
Seven mandates. Or all of them at once.
Each is a discipline with its own vocabulary and its own counterparties. Taken separately, pick the one you are living in. Taken together, they are how a deal gets assembled from parts that would otherwise never find each other.
Data center capacity sourcing
Retail and wholesale colocation, up to 150 kW per rack, sourced globally and negotiated against real market comparables.
Open the mandate → ComputeGPU cluster procurement
H100, H200, B200, B300, GB300, RTX 6000 and RTX 5090 — including allocation reached through channels that never see an open market.
Open the mandate → ContractsGPUaaS & AI cloud negotiation
Reserved capacity, committed-spend agreements and on-demand pricing, read line by line before you sign a multi-year number.
Open the mandate → DevelopmentPowered shell & build-to-suit
Site selection, powered shell terms and build-to-suit structure for buyers who need capacity that does not exist yet.
Open the mandate → EnergyPower capacity procurement
Interconnection position, PPAs and behind-the-meter options. Power is the binding constraint, so we start there.
Open the mandate → AdvisoryBuy-side negotiation advisory
We advise from behind, lead alongside you, or run the table outright on any enterprise infrastructure contract. You decide how visible we are.
Open the mandate →Capacity commercialization
Qualified offtake, anchor tenants and term structure for data centers and neoclouds with capacity to fill — including the KYC and compliance work on the ultimate offtaker, run before any introduction.
Open the mandate →Arrive with the numbers an operator asks first.
Megawatts, density, rack count, cooling class. Work them out in thirty seconds with the estimator — no email address — then read the brief for the platform you are buying.
GB300 NVL72
72 Blackwell Ultra GPUs in one NVLink domain, liquid-cooled and specified to the facility from day one.
Read the brief →Blackwell UltraB300 clusters
HGX B300 allocation routes, density and the terms that matter at cluster scale.
Read the brief →BlackwellB200 clusters
Where HGX B200 fits against B300, and how to contract for it.
Read the brief →HopperH200 clusters
Allocation and reallocation when the previous generation is the right economic answer.
Read the brief →How we are structured, and why it keeps us honest.
Everyone in this market is paid by someone. What matters is whether that changes the advice. Ours is structured so it does not, and the reasons are below.
Alignment
Never paid out of your deal.
Winning business carries a cost of sale that exists whether or not we are involved. That is where we are paid from, so our involvement is funded by the transaction rather than added to it.
Independence
We buy nothing and sell nothing.
No inventory, no listings, no resale position, no stake in what gets signed — so nothing we put in front of you is something we need you to take. What we provide is access: to capacity, to allocation, to power, and to the parties on the other side. You contract directly with the counterparty and own that relationship outright.
Whole-stack assembly
We can bring the deal, not one piece of it.
A complete AI infrastructure deal usually needs three things that live in three different places: an offtaker with real demand, energised capacity that fits it, and compute to fill it. Normally that is three separate searches run by parties who never meet. We hold relationships at every layer, so we can assemble the pieces into one transaction — and every party at the table knows we are working every part of it.
Direct access
Straight to the person who decides.
Relationships at every layer of this market already exist, so a requirement reaches the person who can actually say yes rather than entering a process. Nothing has to be circulated to find them, which is why a mandate can run to completion without your position becoming common knowledge.
Continuity
We stay after the signature.
The engagement does not end at the close. We keep advising as the market moves and bring you what is coming next, so you are early to the opportunity rather than late to it.
Discretion
Confidential by default.
Anonymous at first contact, MNDA on request, and nothing about your position leaves the room.
Why this market is harder than it should be
Too many layers between the people buying and the people building. We remove them.
Most of what passes for market intelligence here is secondhand and sold by someone with a position in the outcome. We work directly, we put the real numbers on the table for whoever we are retained by, and we grow the relationship alongside them rather than sitting between them and it.
Work directly with us →The same four steps, every engagement.
No discovery theatre and no proposal you have to decode. This is the sequence, and you will know which step you are on at all times.
- 01
Qualify
We establish what the requirement actually is, what it is worth, and whether the counterparty can carry it. Nothing moves before this is settled.
- 02
Map
We bring forward the partners in our ecosystem who lead in exactly what you need. Every one of them is vetted, proven, and someone we would put our own name behind.
- 03
Negotiate
We advise from behind, lead alongside you, or run the table outright. You decide how visible we are.
- 04
Execute
We do not hand over a name and step back. We work the problem alongside your team until it is solved, then keep advising as the market shifts and bring you what is coming next before it is widely known.
What a deal clears before it reaches you.
We do the vetting and we do the qualifying, on both sides. Everyone in our ecosystem is a leader in what they do, and you see the shortlist along with the reasoning behind every name on it.
We are only as good as the last deal we put our name on.
- —Counterparty verified
Every partner we introduce is proven in their field before they enter our ecosystem.
- —Capability matched
We bring forward the ones who lead in exactly what you need, not the ones who are merely available.
- —Commercials benchmarked
We know what this market is paying, so you know whether the number in front of you is fair.
- —Accountability held
We stay on the file alongside your team until there is a solution, then stay in the relationship after it. Reputation is the only collateral we have.
Partner spotlight
The cheapest megawatt is the one you already have.
Power at the meter and heat rejection are different constraints with very different price tags. When it is the second one stopping you, the answer is rarely a new lease.
Our partner OptiCool Technologies builds two-phase refrigerant cooling for high-density and AI workloads — over 7,000 rear-door heat exchangers across more than 125 sites, water-free at the rack, and deployable against live racks without taking revenue offline.
For a buyer that can mean density inside a footprint you already pay for. For an operator it means more revenue per square foot on a hall already built, permitted and financed.
Where we run mandates.
Capacity is a local problem with a global market. Power price, permitting speed, latency requirement and tax treatment all move by region, so the right answer is rarely the market nearest you.
Northern Virginia, Dallas–Fort Worth and ERCOT, Phoenix, Atlanta, Chicago, Columbus, plus secondary markets where power is available sooner.
Iceland, Norway and Sweden — low-cost renewable power, free cooling, and the strongest cost-per-megawatt case in Europe for training workloads.
London and Slough for latency-bound inference; Ireland where grid position allows it.
Frankfurt, Amsterdam and Paris for the FLAP-D core, with Nordic and Iberian alternatives when FLAP-D power is constrained.
Singapore, Johor and Malaysia, and Japan — where the Singapore moratorium legacy has pushed capacity across the strait.
Brazil, Chile and Mexico — hydro and renewable-heavy grids with improving connectivity and a favourable cost basis.
Each region fails differently, and the failure mode is what decides your date. Read the market notes for the United States, Canada, EMEA and APAC, or start with how a market gets ruled in or out.
Worth knowing first.
Answered here so the first call can be about your deal instead of about us. The first one is the question we get most.
01You work with buyers and with operators. How does that work?
Often we work all of it at once, and that is the point.
A complete AI infrastructure deal usually needs three things that live in three different places — an offtaker with real demand, energised capacity that fits it, and compute to fill it. Normally those are three separate searches, run by three parties who never meet, on three different timelines. Because we hold relationships at every layer, we can assemble them into a single transaction: bring qualified demand to an operator, bring capacity to a buyer, bring GPUs to either.
Nobody is being played off against anybody. Every party at the table knows we are working every part of it, and we say so at the outset — disclosure is the mechanism, not exclusivity.
What keeps it straight is that we hold no inventory and take no position in the outcome. We do not earn more from one configuration than another, so the only version of a deal we have any reason to push is the one that actually closes and still makes sense in year three.
02Do you work with brokers or resellers?
No. We act for the end user — the party whose workload runs on the hardware and whose balance sheet carries the term — or an advisor that party names in writing.
The principled reason is that it is not what we do. The practical reason is that operators will not quote real numbers into a chain, so nobody in a daisy chain ever sees one. If you are an intermediary who can name and verify the ultimate offtaker, that is a different conversation and we will have it.
03What are we actually hiring you for?
The ecosystem, and the speed it gives you. We already know who leads in each part of this market and how to reach the person who can actually decide, so you spend your time on the right conversations instead of finding them.
04What does this cost?
For buy-side mandates, nothing directly — winning business carries a cost of sale that exists whether or not we are involved, and that is where we are paid from. For operators, we are a small line in your operating cost and we work as an extension of your team, doing the vetting, the qualifying and the negotiation so that what reaches your desk is ready to execute.
05Who pays you, and how does that not distort the advice?
The ecosystem does. We hold no inventory and take no position in the outcome, so nothing we bring forward is something we need anyone to take.
The honest answer to why that holds is not a values statement, it is arithmetic: we are only useful to this market for as long as both sides keep trusting us. One deal where we bent a number to close it costs us every deal after it. Reputation is the only collateral we have.
06Is it too early to talk to you?
No. The earlier we are involved, the more room there is to shape the terms rather than react to them. A first conversation commits you to nothing and often just clarifies what you are actually looking at.
07How fast can you move?
First response within one business day. The relationships are already in place, so we are not starting a search from zero, and for qualified offtake we work from an existing bench of vetted demand.
08Which regions do you cover?
North America, the Nordics, the UK and Ireland, continental Europe including Frankfurt, Amsterdam and Paris, APAC including Singapore, Japan and Malaysia, and Latin America. Full coverage detail is above.
Tell us what is on the table.
Early, mid-negotiation, or just weighing options. There is no stage that is too soon, and the first conversation costs you nothing but the time it takes to have it.
A live requirement?
If you already know the megawatts, the density and the date, start with the offtaker brief — completed properly it secures a position faster than a conversation will.
Operators & data center owners
Have capacity to fill? Skip the form. Send us the specification and we will come back with qualified demand.
sales@thechiefnegotiators.com →