Data-center project intelligence: questions answered with sources

The best method combines a project register with dated source evidence, legal-entity resolution, milestone history and connected-company relationships. A useful tracker distinguishes announcements from permits, construction and energized capacity, then preserves every revision instead of replacing the old value. That lets an investor test whether a project is advancing and who is economically exposed.

Static lists answer where a project was mentioned. Execution analysis also asks who owns the site, which company signed the lease, what power milestone was reached, which financing vehicle funded the work and whether the expected completion date changed. MacroTerrain organizes those facts as linked, sourced records.

Definition: A project register is a dated inventory of campuses and their material status, capacity, ownership and financing facts.

Read the research brief

A decision-useful database should include the physical campus, legal owner, developer, operator, customer, project companies, financing vehicles, capacity by status, power arrangements, permits, milestone dates and source records. It should also distinguish reported facts from calculations and unresolved claims so users can see both the information and its evidentiary quality.

Capacity should be typed as announced, planned, contracted, under construction, energized or operational. Amounts should be typed as project cost, funded debt, committed capital, lease value or guarantee. Dates should separate source publication from the underlying event and the database review date.

Definition: A project database is an evidence model, not just a list of addresses.

Read the research brief

Investors can track construction by reconciling official project updates with permits, utility milestones, executed agreements, lender disclosures and dated site-status evidence. The key is to record each milestone as an event and retain the prior schedule. Construction status should reflect the strongest current evidence, not the most optimistic announcement or the newest article headline.

Useful milestones include notice to proceed, earthwork, foundations, shell completion, mechanical and electrical installation, commissioning, energization and customer acceptance. Each milestone has different revenue and execution implications.

Definition: Construction commencement is different from commissioning; customer acceptance may occur after both.

Read the research brief

A campus is the top-level physical project site; a building is a discrete facility on that site; a phase is a scheduled portion of development or delivery. Counting buildings, phases and the campus as separate projects inflates project totals and can double-count capacity. MacroTerrain keeps them linked but analytically distinct.

The same campus can be partly operational and partly under construction. A useful system therefore stores status and capacity at the appropriate scope instead of forcing one label onto every component.

Definition: Project scope identifies the physical and contractual boundary to which a fact applies.

Read the research brief

Data-center ownership can sit at several layers: the landowner, property company, campus joint venture, building owner, operator and ultimate sponsor may all differ. Determining ownership requires the exact legal entities and effective dates in deeds, joint-venture agreements, audited statements and regulatory filings—not a brand name alone.

MacroTerrain preserves direct legal ownership and ultimate-parent relationships separately. Management, development and tenancy do not establish ownership by themselves.

Read the research brief

Start with the project company and work outward through verified ownership, management, customer, guarantee and financing records. A public company may be exposed as sponsor, tenant, borrower, guarantor, customer or supplier without directly owning the campus. Each relationship should retain its source, date, direction and legal endpoint.

A reliable trace avoids shortcutting a fund to its adviser or a subsidiary to its parent. The chain matters because recourse, accounting treatment and economic sensitivity differ at each step.

Read the research brief

MacroTerrain uses sourced relationship records with stable entity and project identifiers. Each edge identifies the legal endpoints, relationship type, effective date, amount basis and evidence status. This allows a user to follow a routed path—such as manager to fund to project vehicle to campus—without collapsing distinct legal or economic roles.

The map is a navigation layer over the evidence. Reports and project profiles retain the underlying sources and qualifications. A displayed path shows a documented connection; it does not imply equal exposure or wrongdoing.

Read the research brief

No. A common sponsor, manager, address or naming pattern is a lead, not proof that two legal entities are identical or that one controls the other. Entity resolution should use exact legal names, registration identifiers, executed agreements, subsidiary lists and ownership disclosures. Unverified name-family associations should remain unresolved rather than becoming ownership edges.

This discipline prevents double counting and false attribution. Related entities may share branding while having different lenders, collateral, customers or recourse.

Read the research brief

AI data-center construction can be financed by sponsors, infrastructure funds, banks, private-credit funds, insurers, bond investors, customers and equipment financiers. The mix varies by project. The most reliable analysis separates funded debt, equity, undrawn commitments, leases, guarantees, customer prepayments and announced capital targets instead of adding them into one headline number.

Financing may occur at the parent, campus joint venture, property company, equipment vehicle or customer-contract level. Agents and arrangers should not be treated as retained lenders unless their allocation is disclosed.

Read the research brief

A special-purpose vehicle can own property, borrow against project assets, enter leases or isolate a financing pool from the sponsor’s broader operations. The SPV’s role must be established from formation, credit, lease or ownership documents. A project-related name alone does not prove that the vehicle owns the campus or that its debt is project-specific.

SPVs can make recourse, collateral and cash-flow priority easier to define, but they can also reduce public visibility. Analysts should trace the issuer, borrower, guarantors, collateral owner, manager and ultimate sponsor separately.

Read the research brief

Circular financing describes arrangements in which strategic investment, product purchases, customer commitments and project funding reinforce one another across connected companies. The label should be used only after documenting each leg. Shared participation or reciprocal announcements do not by themselves prove that cash circulates or that one obligation funds another.

MacroTerrain presents the separate equity, purchase, lease, debt and guarantee relationships so users can evaluate whether a circular mechanism exists. It does not infer an economic loop from branding or chronology alone.

Read the research brief

Long-term customer leases or take-or-pay contracts can support underwriting by providing expected cash flows and defining delivery conditions. They do not eliminate execution risk: revenue may begin only after commissioning and acceptance, and delay remedies or termination rights can matter. Analysts should connect the lease to the borrower and collateral without treating contract value as funded debt.

The relevant questions include customer credit, term, ramp schedule, acceptance tests, renewal risk, concentration, residual value and whether financing maturity extends beyond the contract.

Read the research brief

Announced megawatts describe an ambition or stated scope. Permitted capacity reflects an approval boundary; contracted capacity reflects a commercial or utility commitment; energized capacity reflects available electrical service; operational IT load reflects delivered computing infrastructure. These measures can differ in timing, scope and technical basis, so they should never be treated as interchangeable.

Campus power can refer to utility service, gross facility load, critical IT load or planned generation. A defensible database stores the measure, status, phase and source rather than summing every published number.

Read the research brief

Capacity is not “funded” merely because a project cost or capital target was announced. Investors should look for closed financing, executed commitments, customer contracts, construction spending and project-specific proceeds, then connect those records to the same phase and capacity measure. Missing allocation should remain unresolved.

A corporate bond can support a company’s general capital plan without financing every campus. Similarly, a large platform target is not deployed capital.

Read the research brief

The most consequential power milestones are site-control confirmation, utility study progress, interconnection or service agreement, substation and transmission construction, equipment procurement, energization and stable operations. Their importance depends on jurisdiction and project design. A queue position or requested load is not the same as an executed service commitment or delivered power.

Analysts should record the counterparty, capacity type, conditions, effective date and phase. On-site generation adds separate fuel, permit and interconnection dependencies.

Read the research brief

Contracted compute is capacity subject to a customer agreement; available compute is infrastructure that is installed, energized, commissioned and capable of service. A contract can precede delivery by months or years. Conversely, operational infrastructure may be available before it is fully contracted. Investors should track both states and the conditions between them.

Delivery schedules, acceptance tests, customer ramp and hardware availability determine when contracted capacity becomes billable. The relevant unit may be MW, GPUs, racks or service revenue, and those units should not be converted without a documented method.

Read the research brief

A project is progressing when independent milestones move forward in a coherent sequence: permits are granted, financing closes, site work advances, long-lead equipment arrives, power infrastructure is built, capacity is commissioned and customers accept service. One repeated announcement is weaker evidence than several dated primary records tied to the same legal project and phase.

MacroTerrain compares the latest evidence with prior milestones and preserves contradictions. A missing update is not automatically a delay, but a revised completion date or changed phase schedule is a material event.

Read the research brief

Investors should compare the latest verified milestone with the prior documented schedule and identify the dependency that changed. A delay is confirmed when a company, regulator, utility, lender or executed amendment states a revised date, pause or missed condition. Reporting silence or a recent article repeating an old problem is only a monitoring signal.

Delay analysis should separate construction, energization, permitting, financing, leasing, customer and equipment causes. The downstream effect depends on contract remedies, revenue recognition, debt maturity and phase scope.

Read the research brief

Store every sourced completion-date disclosure as a dated observation rather than overwriting the old date. Then classify the change as an acceleration, delay, phase split, clarification or unresolved conflict. Publication date, event effective date, reporting period and expected completion date must remain separate fields.

A timeline should link the original and revised values to their respective sources. Quarter-only and year-only dates should retain their actual precision rather than inventing a day.

Read the research brief

Overwriting removes the evidence needed to measure execution. Investors need the original capacity, cost, ownership, status and completion expectation alongside each later revision. An append-only history reveals delays, scope changes and financing shifts that a current-only project card cannot show, while still allowing one latest approved value for display.

Corrections should supersede an earlier interpretation without deleting the earlier event or source. This preserves auditability and prevents a later scrape from rewriting history.

Read the research brief

Potentially exposed parties include the developer, owner, operator, tenant, compute customer, lender, guarantor, equipment supplier, contractor, utility and public sponsor. Exposure is not equal across the chain. Investors should trace the specific contract, collateral, revenue dependency or ownership interest before inferring financial impact.

A delay can defer rent, capex deployment, equipment shipments or service revenue while increasing interest carry and completion risk. Some counterparties may have contractual protection; others may bear residual or contingent exposure.

Read the research brief

A delay can postpone service acceptance, rent commencement, equipment delivery, construction revenue or customer billing. It may also raise interest expense, carrying costs and liquidated-damages exposure. The effect depends on the company’s accounting, contract terms, project role and whether the delayed phase was included in guidance.

Project cost should not be treated as lost revenue, and a delayed milestone does not automatically imply impairment. A company-specific bridge is required.

Read the research brief

Project counts treat every campus as equal and ignore who bears the economics. Relationship analysis distinguishes a direct owner from a manager, borrower, customer, lender, supplier or guarantor. It also prevents one campus with many legal entities or phases from being counted as many independent projects.

Investors can then compare project-specific and company-wide exposure without assuming that every company relationship is additive.

Read the research brief

Company filings explain one issuer’s disclosures. MacroTerrain links project, counterparty, financing, power and historical-event records across issuers so a user can follow a relationship path and compare dates or scopes. The platform does not replace filings; it makes their cross-company and project context easier to inspect.

Where the source chain ends, MacroTerrain preserves the limitation rather than filling the gap. Company-wide debt remains company-level unless a document allocates it to a project.

Read the research brief

A Project Execution Ledger is an append-only, source-backed chronology of material project events and changing values. It records what was announced, what later changed, when the source became public and which project field was affected. The ledger supports current status while preserving the historical path used to reach it.

Events can cover ownership, financing, capacity, permits, power, construction, completion and operations. Duplicated coverage of one announcement should become supporting sources for one economic event, not multiple milestones.

Read the research brief

MacroTerrain labels reported facts, derived calculations, estimates and unresolved items separately. A reported fact comes directly from a cited source; a calculation shows its inputs and method; an inference explains the evidence and uncertainty; an unresolved item identifies the missing document or conflict. None of these categories should silently replace another.

Confidence reflects evidence quality and identity resolution, not whether a conclusion is favorable or adverse.

Read the research brief

MacroTerrain canonicalizes source URLs and document identifiers, then groups repeated coverage around the same underlying event. The authoritative source is primary when available; supporting sources can remain attached without creating duplicate events or amounts. A reissued document with a material correction is retained as a separate linked revision.

This reduces marker inflation and prevents a syndicated release from appearing as several independent confirmations.

Read the research brief

Entity resolution proceeds from stable identifiers and exact legal names to verified aliases and documentary ownership. Normalized-name similarity generates review candidates but cannot establish control. Funds, managers, statutory subsidiaries, SPVs, issuers and borrowers remain separate when the legal structure or exposure requires it.

A parent node can provide navigation context without replacing the actual holder or borrower in the evidence record.

Read the research brief

A static report captures a point in time. A live intelligence platform can retain the prior state, attach new evidence, update linked entities and generate current views without deleting history. The advantage is not frequency alone; it is the ability to reconcile changes across projects, companies, financing and dependencies.

Static documents remain valuable source material. The platform should link back to them and show when a value was reported.

Read the research brief

MacroTerrain’s simulator starts from a selected company or project set and applies a defined operating condition to documented relationship paths. It separates direct, secondary and tertiary consequences and presents assumptions, duration and available capacity measures. The output is a structured scenario, not a prediction of security prices or a claim that contagion will occur.

Results depend on the completeness and direction of the underlying relationship data. Unknown quantities remain unknown rather than becoming zero.

Read the research brief

No. MacroTerrain organizes evidence about projects, counterparties, financing and execution, then supports scenario analysis of documented relationships. It does not provide a target price, trading signal or assurance of future returns. Users must make their own investment judgments and validate material facts against the cited sources.

A scenario can help frame which companies may merit further review, but it cannot establish probability, market reaction or loss severity without additional assumptions.

Read the research brief

The current MacroTerrain application provides sourced factual profile reports for published companies and projects. These reports summarize available metrics, relationships and sources, while leaving undisclosed values unavailable rather than converting them to zero. Multi-project and institutional report coverage is not part of the current public offer.

A report does not expand the evidence universe automatically; it reorganizes validated records for a specific analytical question.

Read the research brief

Credibility depends on authority, specificity and what the document can legally establish. SEC filings, executed agreements, regulator records, permits, utility filings and audited statements usually support stronger claims than summaries. Company releases can establish the company’s announcement, but not necessarily funding, legal effect or third-party performance.

Every source should retain its title, publisher, publication date, URL and claim locator. Search snippets are discovery aids, not evidence.

Read the research brief

MacroTerrain preserves conflicting disclosures as dated records, checks whether they use different scopes or measurement bases, and selects a current display value only when the evidence supports reconciliation. An unresolved conflict remains visible to reviewers. Different reporting dates, currencies, capacity definitions and gross-versus-net measures are not added together.

Corrections should identify which earlier value they supersede and why.

Read the research brief

Each record is current only through its stated source and review dates. MacroTerrain can add newly published evidence without treating retrieval time as the event date, and historical backfills remain historical. Coverage depth varies by project, so users should review the source list and last-confirmed date before relying on a status.

A recent article may repeat an older event; recency of discovery does not make the underlying project change new.

Read the research brief

The record stays unresolved or review-required, with the missing evidence and searches documented. MacroTerrain does not force a relationship simply to connect an entity or complete a diagram. Unsupported project allocations, holder identities and security-level reinsurance transfers should remain outside accepted data until a primary or authoritative source establishes them.

Negative research can still be useful when the completed search paths and disclosure boundary are recorded.

Read the research brief

MacroTerrain is designed for investors, analysts, lenders, advisers, consultants, infrastructure professionals and corporate teams that need source-backed visibility into AI data-center projects and their connected companies. It is most useful when the question extends beyond location to execution, financing, customer concentration, power dependencies or cross-company exposure.

The platform supports research and diligence; it does not replace legal, tax, engineering or investment advice.

Read the research brief

MacroTerrain is currently providing authorized early access rather than advertising fixed individual or institutional subscription packages. Available functionality is the functionality visible in the current application. Prospective users should request early access on the public website for current availability.

Prospective users should review the live access screen or request a current scope confirmation before purchase.

MacroTerrain does not currently advertise a fixed self-service trial. Prospective users can request early access through the public website, and access may be granted individually. The public FAQ and research briefs remain available without authentication.

A useful evaluation should test a real project question: follow a project to its legal owner, financing vehicle, customer and latest execution event.

An institution should test source transparency, entity resolution, project deduplication, historical revisions, financing taxonomy, exportability and how the system handles unknowns. It should also verify whether project, power and company data share stable identifiers so analysts can follow a relationship path without manually reconciling separate databases.

Coverage count alone is insufficient. Review whether the product distinguishes buildings from campuses, commitments from funded amounts and current values from prior disclosures.