Drawdown-risk OS for GCC data-centre lenders that catches covenant drift and delay losses before each construction disbursement.

1. Quantified delay losses turn schedule slippage from an ops annoyance into a finance-line-item that can justify new lender software budget.
2. When incumbent diligence can cost up to $1 million per engagement and still lands as a static PDF, a purpose-built monitoring tool has room to win on both cost and timeliness.
3. Buyers are being trained to expect continuous lender-grade risk intelligence rather than one-off reports, which makes a workflow product around drawdowns and covenant monitoring newly credible.
4. GCC data-centre build-out is concentrated enough that a startup can win a regional beachhead before expanding to other digital-infrastructure lenders.

Catalyst. Azraq's funding plus cited evidence that 57 percent of projects slipped and that a one-month delay on a 60MW campus can cost $14.2 million show data-centre lenders now have a quantified reason to replace static review with continuous monitoring.

The product is a post-close risk operating system for data-centre lenders. It connects sponsor schedules, lender covenants, engineer reports, utility and permitting updates, labour and infrastructure signals, and market-demand assumptions into one deal graph, then produces a live drawdown-readiness score for every project. Instead of waiting for a quarterly PDF, portfolio teams get exception alerts when schedule float, power-delivery milestones, or regulatory conditions move outside approved thresholds, along with an evidence-backed waiver or hold memo. The first workflow is monthly construction disbursement approval; the second is portfolio watchlisting across multiple live campuses. Over time, the company builds a proprietary dataset linking early risk drift to delays, waivers, restructurings, and recoveries across digital-infrastructure loans.

What's different. Consultants sell episodic studies, sponsor-side developer tools optimize site selection, and generic project-controls software stops at schedule tracking. This company owns the lender's moment of truth: the drawdown, waiver, and covenant decision that decides whether capital moves. Its moat grows from mapping live slippage, labour, regulatory, and infrastructure signals to actual credit outcomes across financed campuses, a dataset neither sponsors nor consultants naturally accumulate.

Jobs to be done

flowchart LR
  Buyer[GCC construction lender] --> Pain[Blind drawdown and covenant risk]
  Pain --> Product[Continuous drawdown-risk OS]
  Product --> Outcome[Faster disbursements with fewer delay surprises]

Executive takeaways

• The GCC beachhead is concentrated enough to sell into directly: Saudi Arabia holds more than 65% of upcoming regional capacity, the UAE still leads operational IT power, and GCC project trackers already count 174+ active or planned projects [3][4][5].
• The sharpest pain point is the monthly drawdown decision, where lenders still rely on independent monitoring, draw-package review, and covenant checks while construction analytics show schedule and MEP slippage emerging earlier than static reports usually capture [17][18][19][22][35].
• Generic loan software is not enough because data-centre downside is driven by power, cooling, connectivity, labour, and localisation constraints that sit outside ordinary covenant ticklers [6][7][8][16][31].
• Competition is real but fragmented: Azraq is the closest narrative match, while BankStride, Buildots, nPlan, and incumbent advisers each own only part of the workflow or trust stack [1][19][21][24][17].
• The most defensible version of this company is not “AI decides credit” but “AI assembles the evidence-linked hold/waive memo,” compounding a proprietary dataset of drift signals, lender decisions, and realised outcomes [1][18][20][25][35].

Market definition

This market is software for post-close data-centre construction credit surveillance: ingesting draw packages, schedule evidence, engineering outputs, and external power/connectivity/regulatory signals to produce a lender-ready approve/hold/waive recommendation before capital moves [4][5][7][17][18]. It sits between independent engineer reporting, horizontal construction-loan administration, and construction AI rather than generic sponsor-side project controls alone [19][21][24].

Customer and buyer

The day-to-day user is a portfolio monitoring director, project-finance risk lead, or construction-loan administrator inside a GCC bank or private-credit team; the economic buyer is the head of infrastructure finance or portfolio risk who owns disbursement discipline, data-residency compliance, and downside protection across a small number of very large facilities [7][17][18][19][20][30].

Buying triggers

• A sponsor submits a draw request after schedule, labour, or power assumptions changed since the last engineer update. [17][18][19][22]
• A lender adds new AI or hyperscale construction exposure in Saudi or the UAE and needs portfolio-level watchlisting, not another one-off diligence pack. [9][10][11][15][32]
• Internal audit or compliance pushes for stronger covenant evidence, disbursement traceability, or locally hosted handling of borrower data. [8][16][18][20]

Willingness to pay

Budget is narrow in logo count but real in dollar terms. FWDstart says consultant diligence can reach $1M per engagement, while lenders already pay for recurring draw reviews, site observations, and disbursement control; that supports six-figure annual software ACVs if the product shortens each draw cycle and reduces false approvals. [1][17][18][19]

Category dynamics

Growth signal Market expected to roughly double by 2030

Validation signals

• Khazna has entered Saudi Arabia with plans for up to 200MW of AI-ready capacity in Dammam.
• center3 is targeting 1GW of total capacity by 2030 and tied that build-out to AI and hyperscaler demand.
• AWS has committed more than $5.3B to launch a Saudi infrastructure region.
• Buildots benchmark data shows major MEP systems on 2025 data-centre projects running below required pace, supporting the need for earlier risk detection.

Regulatory & technical constraints

• Borrower data, draw packages, and portfolio documents may need local handling, explicit audit trails, and conservative access controls to satisfy GCC sovereignty and cybersecurity expectations.
• Telecom and building-network standards make connectivity evidence jurisdiction-specific rather than a generic “data room” problem.
• Power, cooling, and labour signals change quickly and require domain-specific interpretation or the platform will over-trigger false alarms.
• Lenders still need independent evidence trails for draw decisions, cost-to-complete judgments, and covenant waivers.

The competitive field is fragmented rather than winner-take-all. Azraq is the closest lender-grade narrative; BankStride owns generic covenant and loan-workflow automation; Buildots and nPlan attack schedule/progress risk; and incumbent independent engineers still own trust at disbursement time [1][19][21][24][17]. The gap is a neutral, data-centre-specific lender cockpit that converts technical drift into an auditable capital-release recommendation [18][20][23][25][35].

Why incumbents do not win by default

• Independent engineers and disbursement advisers. They win trust today because they fit existing draw and audit processes, but their work is episodic, services-heavy, and slower to compound into a reusable cross-portfolio risk dataset.
• Horizontal loan-monitoring platforms. BankStride-class tools automate covenant and document workflows, but they do not win by default because they lack data-centre-specific models for power, cooling, utility milestones, and AI-campus technical drift.
• Construction progress and schedule AI. Buildots and nPlan are strong at identifying schedule or production risk, yet they are optimized for owner and project-controls teams rather than lender waiver, covenant, and disbursement memos.
• Operators and connectivity ecosystems. Khazna, center3, and hyperscalers influence the infrastructure stack, but they are not neutral lender-side systems of record for comparing sponsor evidence against financing conditions.

The strongest version of this company is not a broad data-centre analytics platform; it is a post-close drawdown-risk operating system for Saudi and UAE lenders approving monthly construction disbursements. The first customer is the head of portfolio risk at a GCC bank or private-credit lender with 3-10 live campus loans, where a sponsor draw request arrives after schedule, labour, power, or permitting assumptions have changed since the last engineer report. Research supports urgency because 57% of data-centre projects slipped by at least three months in 2025, a one-month delay on a 60MW facility can cost about $14.2M in lost revenue, and consultant diligence can still cost up to $1M while landing as a static PDF. The MVP should ingest the draw package, independent engineer report, covenant sheet, and external power or permitting signals, then produce an evidence-linked yes, hold, or waive memo and portfolio watchlist without displacing the engineer of record. Go-to-market, pricing, and onboarding should all revolve around one live loan portfolio: founder-led sales, a paid pilot on the next few disbursements, and annual pricing anchored to a lender-team platform fee plus active project monitoring because buyers are paying to protect capital-release decisions. The wedge is strategically attractive because it sits at the exact capital-movement moment where generic loan software, sponsor-side project tools, and services firms each cover only part of the workflow. The main investor concern is that the visible beachhead is small, with modeled SAM of about $7.0M and year-3 SOM of about $1.4M, so the company must prove repeatability in Saudi and the UAE and then extend the same ontology into adjacent digital-infrastructure lenders, insurers, or financed industrial assets. Two disconfirming gaps still matter most: how much of the Saudi and UAE pipeline is actually third-party debt funded, and whether fragmented sponsor and engineer data can be productized without turning onboarding into a services business.

Problem

• Lenders still approve construction draws from engineer PDFs, sponsor decks, and covenant spreadsheets, so material schedule, power, labour, or permitting drift is often discovered after the capital-release decision.
• Because one month of delay on a 60MW facility can erase about $14.2M of revenue and 57% of projects slipped by at least three months in 2025, slow or inaccurate drawdown decisions create lender-sized downside.
• Existing substitutes split the workflow: independent engineers provide trusted observations, generic loan software manages checklists, and sponsor-side construction AI tracks execution, but no system turns those inputs into one auditable approve, hold, or waive recommendation for the lender.

Solution

• Build a lender workbench that ingests the monthly draw package, independent engineer report, covenant sheet, milestone schedule, and external power or permitting signals into one project evidence graph.
• Generate a human-reviewable drawdown-readiness score, exception list, and evidence-linked yes, hold, or waive memo for each disbursement, plus a portfolio watchlist across live campuses.
• Keep humans and certified engineer inputs in the loop from day one with confidence flags, audit logs, local deployment options, and editable recommendations rather than black-box credit automation.

Why we win

• The wedge sits at the exact moment capital moves, where urgency, budget, and measurable ROI are highest and broad analytics tools are weakest.
• An overlay architecture lets the startup coexist with independent engineers, sponsor systems, and generic loan platforms instead of asking conservative lenders to rip out trusted processes.
• Each deployment compounds a unique lender-side dataset linking drift signals, draw decisions, waivers, and realized project outcomes that consultants and sponsor-side tools do not naturally own.

Milestones

flowchart LR
  Wedge[Saudi/UAE lender drawdown wedge] --> MVP[Evidence-linked drawdown memo MVP]
  MVP --> Proof[Faster approvals and earlier exception detection]
  Proof --> Expansion[More lender portfolios and adjacent project-finance assets]

Founding team

Experiment roadmap

Risk assessment

What must be true

• A meaningful share of Saudi and UAE data-centre construction must be third-party debt funded, yielding enough recurring lender drawdown workflows to support the modeled 4-6 account and 8 project-year year-3 base.
• The first 3-5 lenders must accept software as an evidence-linked recommendation layer without requiring autonomous approvals or full replacement of independent engineers.
• The narrow initial artifact set must cover enough of the monthly draw workflow to cut decision time by at least 30% within the first deployments.
• Independent engineers, technical advisers, and lender admins must share data or partner rather than blocking the startup from the trusted evidence trail.
• The data-centre-specific ontology and outcome dataset must transfer into adjacent digital-infrastructure or project-finance segments before the Saudi and UAE beachhead saturates.

Open diligence questions

• What percentage of live Saudi and UAE campuses are financed with third-party debt and monthly lender drawdowns?
• Which fields in a recent draw package, engineer report, and covenant sheet are structured enough for productized ingestion?
• How often do lenders hold or waive draws because of power, labour, permitting, or schedule drift rather than pure budget variance?
• Will the first buyers accept local VPC deployment, or do they require bespoke on-prem architecture?
• Can independent engineers become channel partners, or do they view the product as billable-work displacement?

Model sanity

• Revenue engine. Base-case revenue comes from 6 paying lender accounts by Q4Y3, with blended annual revenue per account rising from $180K in pilots to $300K as two second-project expansions land.
• Must go right. The first three paid lender pilots must convert into repeatable production workflows so ARPU expands without adding a full sales team before seed.
• Model breaks if. If expansion stalls and local deployment stays custom, the downside case leaves only about $0.2M of cash by Q4Y3 and forces an earlier raise.
• Next-round proof. A credible seed story is 5 production lender accounts, 6-8 active project-years under monitoring, two second-project expansions, and a reusable local-VPC deployment pattern reached inside the 30-month raise.

Scenarios

Sensitivity

flowchart LR
  TargetLenders --> PaidPilots
  PaidPilots --> ProductionPortfolios
  ProductionPortfolios --> ProjectExpansion
  ProductionPortfolios --> Revenue
  ProjectExpansion --> Revenue
  Revenue --> GrossProfit
  GrossProfit --> Cash

Flags: The base case still depends on a very small GCC lender buyer universe, so one missed cluster of accounts can move the model materially. · Revenue assumes blended ARPU reaches the top end of the BP range by late Y3 through second-project or second-desk expansion, not just new-logo growth. · Headcount stays flat after the security hire, so any slip toward services-heavy onboarding would require more implementation capacity and pressure gross margin. · The company is still EBITDA negative in Y3, so management would need to start the seed process before it is fully self-funding.

• Trust and liability. Lenders may hesitate to let software influence construction disbursements on multimillion-dollar facilities. Mitigation: Start as an evidence-linked recommendation layer with human sign-off, audit logs, and side-by-side comparison against incumbent engineer workflows.
• Data freshness gaps. Permitting, power-delivery, and labour signals may be patchy or delayed in some GCC jurisdictions or counterparties. Mitigation: Launch in Saudi Arabia and the UAE first, integrate only repeatable data sources, and expose confidence levels plus missing-data alerts instead of hiding uncertainty.
• Incumbent channel conflict. Independent engineers and project-controls vendors could resist a product that threatens their role in lender reporting. Mitigation: Position them as data and review partners, generating lender-facing workflows that still incorporate their certified inputs and create new monitoring revenue for them.