Independent engineering authority for developers entering the data-centre asset class — and for those scaling within it. From your first hyperscale-adjacent mandate to your asset-class playbook, across MENA, Europe and North America.
The franchise, in five lines
What's actually breaking
The first data-centre mandate a firm takes on is rarely lost on a single technical decision. Hyperscale and large-enterprise tenants procure data centres against published technical protocols, executed by audit teams that have done this hundreds of times before — against a developer that has done it once. The asymmetry of preparation is what catches the first-time entrant out. Most failures here are not failures of engineering competence; they are failures of process discipline against a counterparty that has built its entire operating model around process discipline. The market data below is the part of the conversation that opens an investment committee. The four failure-mode entries underneath are how those numbers actually behave inside a project.
≈50%
of US 2026 data-centre capacity delayed or cancelled
TechRadar / Wood Mackenzie, 2025
4–7 yr
grid-connection wait in primary US markets; EMEA hubs sub-1 % vacancy
JLL Global DC Outlook 2026; CBRE IM
$10–50M
cost of retrofitting AI-density after award on a mid-size build
Dell'Oro Group; The Network Installers, 2026
The four entries below are how those numbers behave in the room. Each is a failure mode we have observed across multiple engagements — coded by the project gate at which it concentrates, and accompanied by the operating consequence the developer actually carries when it lands.
Until two years ago, site selection optimised for land cost, fibre, fiscal incentives and planning runway. Today it is none of those things. Power delivery now governs every other variable — utility load letters, substation timeline, water for cooling, climate envelope, grid stability ratings, neighbouring-load contention. The cheap site becomes the unbuildable site within twelve weeks of due diligence, and the developer discovers it after the option has been exercised.
Developers who continue to scope this gate as a real-estate transaction arrive at FEED with a power contract that does not match the tenant's load profile. Recovery costs nine to fifteen months and rarely closes the gap fully. The discipline that catches it earlier is independent technical underwriting on the site itself, before the option period closes — and it is procured for less than the cost of one month of programme slippage.
EoR practices operate within commercial constraints that make pure self-review structurally difficult. The team that drafted the single-line is the same team that will defend it under audit. The senior reviewer is the practice principal whose fee will be reduced by reopening it. The independent reviewer — when one is appointed — usually sits inside the same practice on a different floor, on the same partnership track.
This is not corruption. It is procurement architecture working exactly as designed. What an external reviewer brings is not deeper technical knowledge; it is the absence of that structural constraint. The Top-10 risk register an independent audit produces in two weeks is usually a list the EoR's senior engineers already knew about — with priorities and proposed mitigations they had been unable to surface inside their own firm.
Tender packs are written by business development teams; projects are delivered by site teams. The gap is growing in every market, and is widest where the GC is bidding its first dedicated data-centre mandate. The office presentation is built around past commercial buildings, shell-and-core fit-outs and refit work adjacent to the density requirements being asked for. The delivery team is often assembled after award.
The fix is not to disqualify the contractor — the regional pool is too thin for that posture to scale. The fix is to qualify the delivery proposition. Tender Technical Review reads the submission against the named delivery leads, the named subcontractor chain, the constructability of the actual drawings and the contractor's experience with the specific equipment selections. The output is a risk-adjusted procurement recommendation, not a desktop scoring matrix.
Hyperscale and large-enterprise tenants now read the operations playbook with the same intensity they read the single-line diagram. Four in ten operators globally suffered a major outage caused by human error in the last three years; 85 % traced back to procedure failures or to gaps in the procedures themselves. This is not an operations problem the developer can hand to the operator at handover — the procedure pack is part of what the tenant is buying.
Authored methodology — UDGR for design quality, RHM for risk handling, HCCM for hot-climate cooling, UBI for constructability — is the cleanest route to a procedure pack that audits cleanly. It also has a second-order effect: it makes the asset insurable on terms that do not assume execution risk. That second-order effect is increasingly what the capital partner is actually buying.
Audit rehearsal — the engagement that follows from F.02 and F.04
Hyperscaler tenants do not negotiate the design audit — they execute it. The developer either submits a pack that survives, or accepts a deferred rent date and a revised offtake economics. We rehearse the audit on behalf of the developer, against the named protocol — AWS DDR, Azure SCI, GCP PRR, Oracle ODA, Meta IPR or one of the sovereign-AI standards we hold under NDA — before submission. It is a procurement non-event run as a structured engineering exercise, not a desk-review with a covering memo. The four phases below are the ones we walk through every time.
Design pack is indexed against the named protocol clause by clause. Every drawing, single-line, calculation pack and test case is paired to its protocol reference. Gaps are identified and tagged before any judgement is made on severity.
Deliverable
Variance ledger — fully indexed
Each variance is scored against tenant-audit weight and corrective cost. Top-10 are pulled into the risk register; the remainder are tracked in the long ledger. The output is a procurement-grade prioritisation, not an editorial opinion.
Deliverable
Top-10 risk register
Mitigation pack is co-authored with the Engineer-of-Record under our independent authorship. Design changes are priced and scheduled. The pack is re-submitted as the audit-ready version, signed off by both authors and the developer.
Deliverable
Audit-ready re-submission
A shadow audit is walked through with the tenant's technical team in a structured session — same agenda, same evidence pack, same Q&A discipline. Findings are captured and resolved before the formal audit window opens.
Deliverable
Shadow-audit log + sign-off memo
Decision matrix
This is the most honest way to describe the value. Across the dimensions your capital committee will ask about by name, here is the difference between the status quo and an engagement with Upterix on the asset side. The methodology, the deliverables and the gate cadence are identical regardless of where the engagement starts.
Dimension
Without independent authority
With Upterix on the developer's side
Dimension
Tenant audit readiness
Without
EoR self-check, scored by the team that authored the design — graded against itself.
With Upterix
Independent audit rehearsed against the named hyperscaler protocol before submission, with a closed Top-10 risk register.
Dimension
Lender / IPB review
Without
Reactive responses to lender Q&A under draw-schedule pressure; concessions surfacing one at a time.
With Upterix
IPB-grade design assurance pack prepared before financial close; single-document defence of the technical case.
Dimension
OEM specification
Without
Vendor-led design with single-source equipment exposure; pricing leverage held by the OEM.
With Upterix
Vendor-neutral specifications cleared against three procurable alternatives; pricing leverage retained by the developer.
Dimension
Constructability of EoR drawings
Without
Design clears desk review; site discovers the cost in execution as RFIs and variations.
With Upterix
UBI™ constructability scoring against the actual GC pool before drawings are tendered.
Dimension
Procedure & operations assurance
Without
Operations pack delivered at handover; never audited against the tenant's procedural standard.
With Upterix
Procedure authored against RHM™ and UDGR™, audit-ready at commissioning, signed by the engineering authority.
Lifecycle wayfinder
A developer's project is not a continuous engineering exercise; it is a series of discrete decisions made at four distinct phases. The wayfinder below is the one we use internally to align engagement scope against the asset's actual stage. Most developer relationships enter at Design or at Pre-build, and they tend to extend forward and backward from there as the engagement compounds.
Land option · Site selection · Investment-committee underwriting
The option period is live and the land economics are being negotiated. Power letters, water rights, grid-stability ratings and climate envelopes are scored against the tenant's actual load profile — before the investment committee is asked to exercise. Most of the schedule variance in the entire programme is decided here, in a window measured in weeks. Independent underwriting in this phase tends to cost less than one month of programme slippage and to remove the most expensive single category of late-stage surprise.
Upterix services at this phase
Concept · FEED · Detailed design · UDGR™ sweep
Concept design locks in the capacity plan, the load envelope, the cooling topology and the structural envelope — the most expensive single decision window in the programme. FEED progresses the EoR pack toward IFC. Independent review against the named hyperscaler audit protocol happens at the 60 % mark, with a full UDGR™ sweep on the design and a UBI™ constructability score against the actual GC pool. Tenant-audit issues are surfaced now, when the cost of resolving them is still a drawing change rather than a variation order.
Upterix services at this phase
Bidder qualification · Technical evaluation · Award
Bidders are scored not on the office presentation but on the proposition that will actually deliver: named delivery leads, named subcontractor chain, equipment commitments and the constructability of the bid drawings. Risk-allocation traps in the contract documents are cleared before signature. The output is a risk-adjusted procurement recommendation that the developer's commercial team can take into a final negotiation with both hands free.
Upterix services at this phase
Construction · BIM authority · Audit rehearsal · Commissioning
IFC drawings move into execution. Field clash and RFI cadence are monitored against the UBI™ baseline. The asset-side BIM authority keeps the model federated through handover and prevents the late RFI cascade that destroys closeout margin. At commissioning, the audit rehearsal is walked with the tenant's technical team in a structured session — same agenda, same evidence pack as the formal audit — so the formal session becomes a sign-off rather than a discovery exercise.
Upterix services at this phase
Partnership
A firm that decides to build its own data-centre engineering capability is committing to an eighteen-month team-formation exercise. Three to five senior engineers with hyperscaler track records must be hired in a labour market where every operator and every consultancy is competing for the same names. Proprietary methodology must be authored from scratch and validated through projects that have not yet been won. Vendor relationships that the existing OEM ecosystem will not extend to a newcomer must be built one quote at a time. The cost of that programme — in salary, in time-to-credibility, in attempted-but-not-won bids — typically exceeds the gross margin of the firm's first three data-centre mandates.
The economics rarely work for a firm whose pipeline holds fewer than five DC assets over a three-year horizon. For the firm that does have that pipeline, the eighteen-month delay is a strategic loss the market does not give back. The alternative is to rent rather than build. Upterix is the senior engineering bench, the codified methodology, the audit-rehearsal capability and the hyperscaler-protocol fluency that a firm would otherwise spend two years assembling internally. The firm carries the construction risk, the commercial relationship, the brand and the asset on its balance sheet. We carry the engineering authority and the audit defensibility.
Three depths of partnership
For a firm taking on its first hyperscaler-grade or sovereign-AI mandate, the value of the partnership concentrates in three places — credibility air-cover at procurement, vendor-neutrality at specification, and the audit rehearsal that turns the formal session into a sign-off rather than a discovery exercise. For a firm already in the asset class and planning a portfolio, a new geography or an edge-DC network, the value shifts from market-entry to operational scale: methodology authored once becomes a reusable playbook, geographic playbooks port to adjacent jurisdictions, and the strategic question moves from whether the next project is deliverable to how many of the next ten you can carry in parallel. The three depths below are the structures that house those two journeys.
The entry point for most developer relationships — one named asset, one MSA, full Upterix methodology applied to a single named project. You take on the mandate, we carry the engineering authority behind it.
Entry criteria
First engagement — Rapid Audit conversion or single-engagement advisory on a live asset.
Master MSA across two or more named assets. The methodology authored for the first project becomes a reusable playbook for the second and third — your team starts to recognise its own patterns. You are no longer a developer with a DC project. You are a DC developer with a portfolio.
Entry criteria
Track record of two completed engagements; portfolio pipeline of three or more assets.
Multi-year alliance per asset class. Retained team, joint roadmap, shared methodology cohort. At this depth the engineering authority is no longer a vendor — it is a standing feature of how your firm operates in the asset class. One or two anchor partnerships per category, per geography.
Entry criteria
Board-level conversation on both sides — typically one to two anchor partnerships per asset class per geography.
The boundary — what we are not, on purpose
Partnership requires precision about boundary. We are not the Engineer-of-Record on your assets — that responsibility belongs to a registered EoR firm in each jurisdiction. We are not the General Contractor, and we do not bid against the firms we engage with. We do not act as procurement agent and we do not hold OEM commissions, exclusive distribution rights or vendor equity. We do not carry construction risk on our own balance sheet. Each of these constraints is what allows the engineering authority we provide to be worth what we charge for it.
Anonymised risks we have closed
The cost-avoided figures below are the developer's own internal accounting, not a marketing estimate. The pattern is more useful than any single record: the same handful of failure modes, caught at the gate where they originated, at a fraction of the cost they would have carried at handover. Across these three engagements alone, the cost-avoided total is north of ten million dollars.
caught at · Site Selection & Due Diligence
caught at · Design Review & Technical Risk Assessment
caught at · Tender Technical Review
Hyperscaler-audit readiness self-check
Run this against the asset you are currently underwriting. No data is sent anywhere — it computes locally in your browser. If your score is below four, your offtake date is at material risk. The recommendation below the gauge is the same one we would give a developer on a Discovery call.
Diagnostic — six questions
Your answers stay on this device. The score updates live as you go.
Has the EoR design pack been mapped clause-by-clause against the named tenant's audit protocol?
Why it matters · The protocol is published or contractually defined. A clause-by-clause map is the only way to know what will be flagged.
Do you hold a Top-10 risk register signed by an independent engineering authority — not the EoR?
Why it matters · An EoR-authored register grades the EoR's own work. Capital partners increasingly require independence.
Is your commissioning procedure pack authored against a named published methodology (UDGR™, RHM™, HCCM™ or equivalent)?
Why it matters · Hyperscaler audits review procedure quality, not just design. Authored methodology is what the audit will be graded against.
Have you walked a shadow audit with the tenant's technical team before the formal session?
Why it matters · Tenant audit teams will not negotiate findings in the formal window. They will discuss them in the dry-run window.
Is your operations playbook audit-ready against the tenant's procedural standard at handover-minus-90-days?
Why it matters · The operations pack is part of what the tenant is buying. If it is not ready, rent commencement defers.
Has your design-assurance position been documented in IPB form for your construction lender or insurer?
Why it matters · Lenders and insurers price the gap between design intent and execution risk. Documentation closes the gap.
Your readiness score
Awaiting answers
Recommendation
Answer at least one question to see a tailored recommendation. The score and recommendation update live.
Questions developers ask before signing
These are the questions that come up in every Tier 2 negotiation and most Tier 1 ones. The answers below are the same answers we give in the room, written down so they are easier to forward.
No — never on the same project, on either side. On parallel engagements where we support a developer on one asset and audit the same tenant on a different asset, a documented Chinese-wall protocol applies: distinct partner-led teams, partitioned document repositories, signed conflict declarations on both engagements, monthly compliance review by an independent third partner. Independence is the franchise — we defend it operationally, not just in marketing copy.
Two engagement teams under two different partner leads, two partitioned document spaces, two signed conflict declarations and a monthly written compliance review by a third partner not on either engagement. The mechanism is audit-able by either client at any point. We will provide the protocol document under NDA on request.
Yes — when both parties consent in writing, and when the scope is structured to keep our opinions independent of either side's commercial position. This is now the most common configuration on sovereign-AI financings, where the developer and the lender share an interest in the same audit outcome and want a single technical voice.
Fee split across project gates — typically 25 % at concept sign-off, 25 % at IFC, 25 % at commissioning rehearsal, 25 % at handover. On programme engagements a per-MW success fee can be layered on top, payable on offtake signature at the modelled rent. The structure is designed to align our P&L with yours, not to maximise hours billed.
Yes. Site Selection & Due Diligence is built for the pre-option-exercise window. The deliverable is structured to feed the investment committee directly — power buildability, water rights, climate envelope, grid stability, planning runway, all scored with documented methodology. Typical scope is four to eight weeks.
AWS DDR, Microsoft Azure SCI, Google Cloud PRR, Oracle ODA, Meta IPR, plus two sovereign-AI compute standards under named NDAs. We do not publicly disclose the sovereign-AI protocols; under engagement NDA we will walk you through exactly what we model and how.
Three ways developers start the conversation
Most engagements begin with a single named project. A smaller share open at the programme level, where the same methodology is applied across a portfolio. The third lane is for capital partners that need independent technical air-cover before they sign.
Lane A — Live project gate
Two-week scoped engagement on a single project. Produces a Top-10 risk register and a costed mitigation plan. Designed as a procurement non-event and structured to convert into a fuller engagement on the same asset.
$5,000 fixed · 2 weeks · Top-10 risk register
Book a Rapid AuditLane B — Portfolio programme
Programmatic engagement across two or more sites under a single MSA. Gate-based pricing, shared methodology, one set of deliverables for the capital committee. Typical scope: 18–30 months per asset.
MSA-based · gate-priced · per-MW success-fee optional
Programme Discovery callLane C — Capital-side advisory
Independent Lender's Technical Advisor opinions, IPB reports and design-assurance briefs for senior debt, project-finance and underwriting committees. We do not hold construction risk on our balance sheet — that is the point.
Scoped per asset · NDA-first · capital-committee ready
Lender's brief