Dedicated Compute.Your Cluster.Your Contract.

Each engagement is structured through an independently formed Delaware LLC (SPV) — for example, Verde Compute Phase 1 LLC. Your cluster, your MSA, and your data rights are held within that SPV, legally isolated from Verde Compute, Inc. and from all other client engagements.

If Verde Compute, Inc. were to undergo any corporate change, your SPV continues operating under its own independent legal structure and the full pre-paid insurance consortium. The 21-policy insurance stack — pre-paid for the entire contract term at Financial Close — remains active regardless of Verde's corporate status. You are also named as Additional Insured on key policies, giving you direct rights with the insurers, not just a contractual claim against Verde.

Your cluster runs on a dedicated, current-generation GPU cluster architecture — Verde's minimum hardware standard for all engagements. Hardware generation and procurement date are documented in the MSA Hardware Schedule at signing — before you commit. Most dedicated infrastructure providers do not disclose hardware generation until deployment, or at all. Verde discloses at signing so you can assess the technology lifecycle before the engagement begins.

Each Verde engagement is a dedicated, single-client deployment. No shared tenancy. No multi-tenant pools. No noisy neighbors. Your GPU allocation is yours by contract for the full term, isolated at both the hardware and network level. Another Verde client cannot run workloads on your cluster under any circumstance.

Target timeline is approximately 120 days from MSA execution to cluster go-live. This covers: vendor allocation confirmation, data center provisioning, hardware delivery and rack installation, liquid-cooling commissioning, GPU stack configuration, and Verde NOC readiness certification.

Hardware delivery is the critical path. The 120-day target is based on documented lead times as part of Verde's engagement planning process. Following MSA execution, Verde assembles your dedicated engineering team. Your named roster — individual names, role assignments, and GPU Infrastructure Certifications — is delivered to you no later than 30 days before your go-live date. This commitment is written into MSA Exhibit A.

If schedule certainty is critical, Verde offers an optional Temporal SLA: contractually defined delivery milestones. Confirmed milestone delays attributable to Verde trigger a documented compensation event, independently verified by the Independent Engineer, providing financial recourse if target dates are missed.

As a Verde client, you are named as Additional Insured on key policies — meaning you have direct rights under the insurance contracts, not just a contractual claim against Verde. The consortium covers:

Hardware All-Risk — full replacement value, new-for-old; no degraded capacity after failure. Business Interruption — if the primary cluster is unavailable, Verde activates emergency cloud compute at Verde's cost; your workloads continue, billed at your standard rate. Cyber Liability — incident response, AI model reconstruction, regulatory fines. Professional Liability (E&O) — Verde's operational errors covered financially. All 21 policies are pre-paid for the full contract term from Verde's financing at close.

Verde's named engineers operate exclusively at the infrastructure layer: 24/7 GPU health monitoring, CDU liquid cooling management, high-bandwidth interconnect and network operations, GPU stack driver and firmware management, SLA compliance tracking, and emergency cloud burst activation.

Model weights and GPU compute: protected by Hardware-enforced Confidential Computing — a Trusted Execution Environment (TEE) that encrypts computation inside the GPU. Verde's monitoring systems receive hardware telemetry only; compute processes are architecturally inaccessible, even to Verde engineers.

Training data and storage: protected by client-managed BYOK encryption keys. Your organization retains sole custody; Verde cannot decrypt storage contents under any circumstance.

OS-layer access: all sessions are logged, time-limited, and require dual authorization via PAM. Your security team receives real-time access logs — every Verde engineer login, every action, timestamped. IP ownership: 100% yours, MSA guaranteed.

Yes. Verde's engagement is structured as a multi-year fixed OpEx commitment. Your organization pays zero CapEx: hardware, installation, GPU platform software stack licenses, vendor SLA, data center colocation, power and electricity, and all 21 insurance policies are pre-financed by Verde at project close and pre-paid for the full contract term.

Your obligation is a single, fixed monthly Take-or-Pay fee. No variable billing. No utilization overages. No power surcharges. No surprise invoices mid-contract. And because Take-or-Pay eliminates any conflict of interest on Verde's side, Verde delivers a monthly FinOps report — GPU cluster utilisation by cost centre in CFO-ready format — so your finance team has full visibility into what the infrastructure is doing, regardless of utilisation level. Indicative pricing range is provided under NDA during engagement.

Yes — and this is by design. Take-or-Pay means your organization commits to a fixed monthly payment for dedicated capacity, whether or not that capacity is fully utilized in a given month. In exchange, Verde commits to making 100% of your contracted cluster available 24/7, backed by an industry-standard uptime SLA from Tier III certified Pacific Northwest data center infrastructure — powered predominantly by hydroelectric energy.

You are not paying for usage. You are paying for guaranteed, dedicated, sovereign AI infrastructure that is legally yours for the contract term. This is the structural difference between Verde and shared-cloud GPU providers.

A Verde engagement is a committed, irrevocable obligation for the full contract term. This is the structural foundation that enables Verde to pre-finance the entire infrastructure — dedicated hardware, named engineering team, and a pre-paid insurance stack — and deliver it with zero client capital outlay.

Verde's insurance architecture is specifically designed so that service interruption is not a permanent state — it is a managed condition. Emergency cloud compute activates beyond the SLA remedy period, funded by Verde's Business Interruption insurance. The service does not stop; it migrates or restores — at Verde's cost, never yours.

Client-initiated termination outside of a verified, uncured Verde contractual breach does not eliminate the remaining payment obligation. The committed payment schedule remains in effect for the full contract term.

Verde's MSA Exhibit A defines the team structure committed to your engagement: role titles, minimum qualifications, seniority requirements, and on-call coverage obligations. Verde's minimum standard across all technical roles is L3+ — NOC Leads, Senior GPU/ML Engineers, Site Reliability Engineers. No L1 or L2 engineers in client-facing roles.

Individual names, GPU Infrastructure Certifications, and background check documentation are compiled into a named Team Roster and presented for client approval prior to go-live. All team members are L3+ GPU Infrastructure Certified before go-live — a contractual requirement, not a target. If a named engineer departs, Verde is obligated to notify the client and provide a replacement of equivalent qualification within the MSA-specified timeframe.

180 days before expiry: Verde commissions an independent hardware valuation at Verde's cost and delivers the report to the client. Hardware generation and procurement date were already disclosed at MSA signing — so there are no surprises at renewal about the asset you are evaluating.

By day 90 before expiry: the client provides written notice of their elected path. Options: (1) renewal on updated terms; (2) hardware acquisition at independently assessed market value — Verde continues full operational handover support for 3 months at no charge; (3) upgrade to next-generation hardware architecture within a new SPV — this upgrade path is defined contractually at signing, not negotiated under time pressure at term-end; or (4) orderly wind-down with full data migration support. Verde has no lock-in beyond the original MSA term. Your data, model weights, and all IP remain yours throughout.

Each Verde engagement is structured through its own dedicated Delaware Special Purpose Vehicle (SPV). Verde Compute, Inc. carries no cross-engagement client liability. The SPV formed for your engagement is a legally independent entity — its obligations, assets, and liabilities are entirely separate from Verde Compute, Inc. and from every other client SPV.

From your organization's perspective: you have a contract with a dedicated legal entity — not a contract line buried inside a larger corporate balance sheet. All client payments flow into a segregated escrow held by an independent institutional trustee — Verde cannot access these funds unilaterally. Verde receives distributions only after all engagement obligations are fully met. This SPV is backed by a full Corporate Guarantee from Verde Compute, Inc. — institutional backing at the parent level. Full capital protection architecture is provided under NDA during engagement.

Yes — Verde operates on 100% renewable energy, sourced from Pacific Northwest hydroelectric baseload and certified through WREGIS Renewable Energy Certificates. This is not a carbon offset purchase; it is a verifiable, traceable energy source.

Verde delivers a quarterly Scope 2 carbon report — WREGIS certified, audit-ready for CSRD, ISSB S1/S2, California SB 253, and equivalent regulatory frameworks. The report includes facility PUE disclosure and renewable energy certificate verification. If your organization has ESG reporting obligations for AI infrastructure, Verde's quarterly report can be submitted directly to your sustainability auditors — no self-assembly of emission factors required.

Full detail on Verde's ESG architecture — including facility certification, reporting schedule, and MSA Environmental Schedule — is available on the Infrastructure page; supporting documentation is provided under NDA during engagement.

No. Verde's MSA contains no "best efforts," "commercially reasonable efforts," or capacity-constraint escape clauses for availability or performance. These clauses — common in standard cloud and GPU rental agreements — legally permit the provider to fall short of stated commitments when conditions are "reasonable." Verde does not use them.

Your cluster capacity is a contractually committed, absolute obligation. If Verde cannot deliver the committed capacity, the Surety Bond — bound at Financial Close — provides direct financial remedy to your organization. This is not a service credit; it is a financial guarantee from a third-party bond issuer, payable to you directly. Verde's model eliminates the legal mechanism that allows most infrastructure providers to avoid liability for underperformance.

Every Verde cluster deployment undergoes a 72-hour full-capacity burn-in followed by a 24-hour on-site acceptance test conducted prior to client go-live. The pass threshold is 97% of factory benchmark performance — a contractually defined standard, not a self-certification.

Acceptance test results are formally certified by the Independent Engineer (IE) — the same third-party certifier that signs off on each construction milestone — and delivered to the client before the SLA clock begins. If the cluster fails the 97% threshold, Verde is obligated to complete vendor remediation within 48 hours before go-live is declared. Industry standard for dedicated GPU infrastructure is either no defined acceptance threshold, or self-certification by the provider. Verde's IE-certified acceptance test means you receive formal third-party confirmation of hardware performance before your organization's production workloads begin.

Yes — Verde offers a Pre-Flight Risk Rating before any commercial commitment is required. Verde's infrastructure team evaluates your AI workload across three dimensions at no charge: (1) Compute Requirement Audit — cluster sizing, memory bandwidth, and interconnect topology against your model architecture; (2) Data Pipeline Readiness — ingestion structure, preprocessing architecture, and storage I/O profile; (3) Infrastructure Fit Rating — a formal A-through-D determination delivered as a written certificate. Rating A or B: Verde can serve your workload as specified. Rating C: specific conditions apply, documented in the subsequent proposal. Rating D: Verde declines under current workload conditions.

Separately, if your production AI deployment is planned 12 to 18 months ahead, Verde offers a structured Compute Option — a signed capacity reservation at today's infrastructure pricing, secured with a deposit substantially smaller than your full engagement commitment. If dedicated GPU infrastructure costs increase before your start date, your locked price is protected. If you choose not to proceed, you forfeit only the deposit. Your full engagement budget is never at risk from a price movement before your engagement begins. The Pre-Flight Certificate and Compute Option Term Sheet are available before any engagement fee is incurred.

Shadow AI refers to AI tools used by employees outside of organizational governance — consumer-facing AI assistants, browser-based tools, and API access to foundation models that have not been reviewed or approved by the organization's IT or security function. 2026 enterprise research consistently finds that fewer than 5% of organizations have full visibility into AI tool usage across their workforce, and that unsanctioned AI tools were a contributing factor in a significant share of breach incidents among affected organizations. This has moved Shadow AI to the top of enterprise CISO risk lists — above phishing and ransomware in a number of 2026 surveys.

For organizations deploying proprietary AI on Verde's dedicated infrastructure, the boundary between sanctioned AI (your models, running on your dedicated cluster) and unsanctioned AI (consumer tools) must be architecturally documented — not merely stated in an acceptable use policy. Verde provides a Shadow AI Firewall Certificate documenting the technical controls that prevent Verde-hosted models from receiving unsanctioned data ingestion: network-layer isolation, VLAN segmentation, and ingress traffic classification logs. This certificate is formatted for submission to your CISO governance programme and increasingly for cyber insurance underwriters who require documented AI perimeter controls as a condition of coverage.

California AB 2013 — the Generative AI Training Data Transparency Act — came into effect on January 1, 2026, requiring enterprises that develop or substantially modify generative AI systems for public use to publicly disclose training data sources, types, volume, intellectual property status, and any use of personal information. For organizations using Verde's infrastructure to train or fine-tune AI models, Verde provides a Training Data Provenance Seal: a timestamped, cryptographically signed infrastructure-layer record of data ingestion events — when data entered the cluster, from what verified source endpoint, and under what access controls.

Verde does not inspect, read, or classify the content of client training data. The Provenance Seal covers the infrastructure chain of custody: source endpoint, transit encryption confirmation, ingress timestamp, and storage location — not content classification. The Seal is structured for submission to legal teams preparing AB 2013 disclosures, not as a replacement for client-side content governance. Clients retain full responsibility for the content and IP status of their training data; Verde's infrastructure-layer documentation supports the client's own compliance workflow.

When a production AI model is deployed on Verde's infrastructure, Verde's Independent Engineer executes a standardised inference test suite immediately at deployment — outputs are logged, timestamped, and cryptographically hashed, and the IE signs the record. This creates a Post-Deployment Behavioral Baseline certificate: an independently verifiable document confirming what outputs the model produced under defined conditions at a specific timestamp. It is not Verde self-certification — it is a signed IE record.

Federal Reserve SR 11-7 model risk management guidance requires financial institutions to maintain documented, independently verified model performance baselines at the time of deployment. For financial services organizations deploying AI models on Verde infrastructure, Verde's Behavioral Baseline certificate is structured to support this requirement at the infrastructure layer — without exposing model architecture, training methodology, or data content to Verde. BYOK encryption and Hardware-Native TEE ensure Verde has zero access to model content. The Baseline covers inference output behavior only. This is a governance document, not a performance guarantee.

Verde assembles a structured AI Governance Documentation Package for each client at engagement initiation and updates it quarterly. It includes: Sovereignty Control Matrix — a documented mapping of five control layers and their respective owners (client, Verde, facility operator), confirming which layers Verde can access and which are architecturally excluded; Independent Engineer Audit Opinions — IE-signed infrastructure performance opinions at Go-Live, annually, and post-incident; Flight Recorder Archive access rights — immutable, tamper-proof operational telemetry retained for a minimum of 7 years, client-accessible in full; Hardware Provenance Passport — OEM-source and supply chain attestation documentation; and Post-Deployment Behavioral Baseline certificate.

This package is specifically formatted for submission to enterprise AI governance teams, external auditors, board risk committees, and regulators. It does not require any representation from Verde's commercial team — every document in the package is either an independent third-party opinion (IE), a factual infrastructure record (Flight Recorder), or a signed client-controlled document (Sovereignty Matrix). Full AI Governance Documentation Package available under NDA.