Returns, overruns, warranty and credits — all of it written here

Arriving at our warehouse within 14 days of the term ending counts as on time. From day 15 we charge 150% of the daily rate, capped at 30 days of that daily rate for the batch. The tracking number issued on the day you ship is the proof of return date, not the delivery signature. The cap only caps overdue rent: still not shipped by day 45 is treated as loss — see the next item.

Charged at replacement value, not capped at the deposit. The deposit covers overdue rent and minor damage; the hardware is worth far more. With no tracking number by day 45 after the return window closes, it is treated as lost and charged at the batch's prevailing replacement value, the deposit applied first and the balance settled separately. Damage outside normal use — drops, liquid, disassembly, non-OEM additions, overclocking — is charged at the repair quote or replacement value, and we send you the quote before anyone touches it. While at your site or in transit, you insure at replacement value and give us a copy of the policy before dispatch — including the return leg if you arrange it. While in our racks it is insured by us and is not your concern.

Yes. 14 days' written notice; days used are settled at the actual rate and the unused portion is refunded in full. If a term past 90 days had moved to tiered pricing, early exit is recalculated at the original rate and the difference invoiced. Two things up front: the deposit is returned after inspection, not early on the notice; and rent is a usage fee — it does not offset a later purchase price.

We replace rather than repair. In our racks: swapped from the local spares pool and power-on tested within 4 hours of the ticket. At your facility: a replacement machine dispatched within 4 hours, then arriving at the pace of freight and customs — that leg is outside our control, so we do not commit to arrival hours. We collect the failed unit. Downtime from hardware failure is deducted from the invoice hourly with no claim needed; the deduction is capped at the rent, and the overall cap is in "Can we claim further after a credit".

Three years on the complete machine, covering GPUs, mainboard, PSUs, fans and backplane. Scope and duration rest on the OEM or ODM support for that batch, with local part swaps executed by us on top. Not covered: failures after customer disassembly or non-OEM parts, damage from non-data-centre conditions (intake above 35 °C or dust-laden), physical damage from transport outside the specified packaging, silicon degradation from overclocking, and consumables and software themselves (drivers, firmware, OS). The warranty follows the serial number; notify us in writing before transfer and it continues once the recipient accepts the same terms — transfers we were not told about are not honoured. A100 comes through second-hand channels, so its warranty is 1 year, or 6 months for used units.

Send an email to quote@apexbyte.ai with the serial number and the symptom — no form. Within one business day you get an RMA number and a shipping address. Inside warranty we start with a part swap: the spare ships within 4 hours, outbound freight inside the EU is on us, outbound freight outside the EU is itemised on the quote, and you fit it yourself (we swap parts, we do not come to site). Only what a part swap cannot fix goes to depot repair: 5 – 10 business days, counted from the day we receive the machine. Past 10 business days we replace it with the same model or one no lower in specification; if the model is discontinued or export controls block a replacement, we refund at prevailing residual value. For failures outside warranty (see above), freight both ways and the repair are yours, and we send you the repair quote before anyone starts.

Four sets of raw data from a 72-hour full-load test: GPU and memory temperature curves (1-minute sampling), correctable and uncorrectable memory ECC error counts, measured all-to-all NVLink bandwidth, and 8-GPU all-reduce throughput. A machine short on any of them does not ship. The PDF travels in the box and is archived against the serial number for 5 years — ask for it again any time.

On monthly uptime: below 99.9% credits 10% of that month's hosting fee, below 99.5% credits 30%, below 99.0% credits 100%. The measurement is actual minutes of power loss at the rack PDU. Four exclusions, stated plainly: your own OS, drivers and application failures; hardware failures in your own equipment; planned maintenance we announced 7 days ahead; and force majeure or large-scale upstream carrier outages (see "Can we claim further after a credit"). A utility power cut is not an exclusion — that is exactly what the UPS and the generator are for, and if they do not hold, we credit. Credits are capped at that month's hosting fee for that rack and are applied automatically to the following month's invoice; you do not file a claim.

Yes. Three requirements: standard 19″ rack-mount, no more than 12 kW per machine, and IPMI or equivalent out-of-band management. We take power, cooling, network, racking and on-site part fitting; we do not take your hardware warranty and are not responsible for its failure rate or lifespan — you supply spares, we fit them within 4 hours of arrival. Before racking we photograph and both sign off the arrival condition, and that record is the baseline: subsequent natural failure and ageing are outside our scope, while physical damage from our handling is credited. Your hardware must also meet export-control and compliance requirements, and you warrant title and provenance.

Three things: machines, space, operations. On-demand cloud sells you a slice of time on someone else's machine — you can stop any time, but past three months of continuous running the unit cost is usually the highest available, and you control neither quota nor neighbours. Here the machine is either an asset you bought or a complete system rented to you by the term, running in a rack in Amsterdam, Frankfurt or Singapore. Power, cooling, uplink, spares and 7 × 24 staffing are ours, with uptime and swap response written into the contract.

Three commitments are in the terms: monthly uptime (below 99.9% credits 10% of that month's hosting fee, below 99.5% credits 30%, below 99.0% credits 100%), a part fitted and power-on tested within 4 hours of the spare arriving, and one monthly report carrying minutes of power loss and the intake temperature curve. What is not committed is written just as plainly: how fast your workload runs, how well your model trains, end-to-end network latency (we commit only to the rack-to-uplink-port segment), interruptions caused by upstream carriers or force majeure, and any form of return or yield — whether this pays off depends on your own utilisation. The credit cap and sole remedy are in the next item.

No — the credit is the sole remedy. The amount calculated above is our entire liability for that event, and cumulatively it will not exceed what you actually paid us in the 12 months before it. We do not carry indirect loss: the compute cost of re-running a training job, lost revenue or profit, lost data or model weights, or third-party claims against you. Data backup is outside our scope — keep your own copies. Force majeure — war, strikes, earthquake, flood, regional grid collapse, government order, large-scale upstream carrier outage — is excluded from the uptime measurement and generates no credit; we notify you in writing within 24 hours, and beyond 30 days either side may terminate without liability. This item applies equally to all three lines.

Yes, and this is not something we route around. High-end GPUs are controlled items, and every delivery is conditional on obtaining the necessary export, re-export and transit licences. At order you sign an end-user and end-use declaration and undertake not to re-export, resell or sub-lease the hardware or the compute to restricted countries, regions or listed entities, and not to apply it to prohibited uses. Where controls, a refused licence or a policy change make delivery impossible, we refund any prepayment in full and neither side pursues the other; penalties and losses caused by a false declaration on your side are yours. Asia-Pacific deliveries typically wait 5 – 15 business days longer for licensing than deliveries inside the EU, and that wait is not counted in the lead time.

The plan comes within one business day. After that: rental ships 2 – 5 days after confirmation; purchase runs 4 – 21 days by configuration, including 72 hours at full load before it leaves; hosting is racked 5 – 10 days after signature. The start points are written into the contract: hosting counts from signature and hardware arrival, purchase and rental from cleared deposit and confirmed configuration. Customs waiting, export licence waiting, and waiting on your facility or network are not counted. You do not need to negotiate a rack contract or wait on a power upgrade first — that part is already done.

DAP by default: we deliver to the address you name, and duty and import VAT are yours. If you want DDP we handle it for 2.5% of goods value, itemised on the quote. Everything is insured at 110% of goods value in transit. Transit times are the carrier's reference figures: 3 – 5 days inside the EU, 7 – 12 days from the EU to Asia-Pacific, both counted from hand-off to the carrier. Customs inspection and export licence waiting are excluded and are not grounds for a delay credit.

Every NVMe and SSD gets a NIST SP 800-88 Purge-level erase, with an erasure certificate carrying the serial number. Erasure is completed within 3 business days of the unit arriving at our warehouse. You can also keep the drives: they are priced by capacity and model, that cost is deducted from the deposit and they are not collected, with the unit price itemised on the quote. The certificate only attests that we performed the step; keeping your own copies while the hardware is still with you, and wiping it yourself before return, are both worth doing.