Frequently Asked Questions

With traditional hyperscalers or GPU clouds, every dollar paid is 100% ephemeral operational cost that yields zero retained asset value once you shut down the instance. Under our lease-to-own schedule, you are assigned a dedicated, serial-tracked NVIDIA DGX Spark node. A major portion of every monthly payment builds equity credits toward outright title ownership at the end of the term.

Yes, absolutely. Our contracts feature a formulaic, pre-calculated early buyout schedule: Buyout = Remaining Principal − Cumulative Prepaid Equity Credits + Settlement Administration Fee. You can exercise this buyout at any time to immediately obtain full title and ownership.

Upon final payment, legal title and bill of sale are issued to your entity. You then have two options:

• Retain in Colo: Keep the machine running in our Burbank–LA cage on low-cost ongoing colocation rates (covering power draw and network port only).
• Ship Hardware: We unrack, package, and ship the chassis directly to your company facility, lab, or home office via insured carrier.

Spark Lease does not provide accounting, financial, or tax advice. Whether a lease-to-own structure is treated as an operating lease expense or capitalized equipment depreciation depends on your business structure and IRS Section 179 rules. Please consult your corporate CPA.

The nodes are hosted in a professional enterprise colocation facility in the Burbank–LA metro area. Due to strict data center compliance, insurance regulations, and physical cage security, unescorted physical visits are not permitted. All access is performed remotely over secure network tunnels, or via escorted appointments if pre-arranged.

No. During the active lease period, the equipment remains secured in the operator’s locked colocation cabinet in Burbank–LA. Physical release and shipping occur only after complete ownership title conveyance.

All units are backed by NVIDIA manufacturer hardware warranty. In the event of a verified hardware malfunction (e.g. SSD, power supply, or GB10 board issue), our SLA covers triage and RMA replacement. If a replacement unit or spare node is available in the cage, workload migration will be coordinated per the service level agreement.

Because the DGX Spark contains advanced NVIDIA Blackwell computing silicon, US Export Administration Regulations (EAR) and Department of Commerce guidelines strictly apply. All lessees must complete standard identity verification (KYC) and confirm that compute resources will not be transferred to embargoed entities or restricted destinations.

Nodes may be used for any lawful machine learning, AI model fine-tuning, inference serving, robotics simulation, or data engineering workloads. Crypto-mining that degrades hardware longevity, denial-of-service activities, unauthorized penetration attacks, and illegal content distribution are strictly prohibited.