NEWVectors or files. Pick a path.Start →

    Specifications

    Every component, what it is for, and how it extends

    The full Mixpeek Appliance build sheet by phase. Each row says what the part does, where it sits, and what it grows into, so nothing is bought twice.

    These are planning ranges, not quotes. Hardware pricing, availability and qualification status all move. Validate the seller, the warranty, the qualification status and the interface before buying anything, and treat totals as a way to size a budget rather than as an offer.

    Where the budget goes

    Each bar sums the low end of the priced line items in that phase, so it is a floor rather than a forecast. The full table is below, unchanged.

    Phase 0, nowfrom $5,400 · 2 of 4 line items priced

    Largest: Storage Pod 01: 4U 24-bay ($4,000) · Seagate Exos X24 24 TB SATA HDD x4 ($1,400)

    Phase 1, V1 rackfrom $31,100 · 19 of 20 line items priced

    Largest: RTX PRO 6000 Blackwell 96 GB ($13,000) · Threadripper PRO 9975WX ($4,000) · + 8 × 20-24 TB enterprise SATA HDDs ($2,800)

    Phase 2, scale outfrom $5,000 · 1 of 3 line items priced

    Largest: Lustre shared hot tier ($5,000)

    Phase 3, frontierfrom $50,000 · 1 of 2 line items priced

    Largest: Dense Seagate-class JBOD/storage shelf ($50,000)

    Off-grid powerfrom $43,000 · 3 of 4 line items priced

    Largest: 48 V LiFePO4 bank ($20,000) · PNW solar array ($15,000) · Hybrid inverter + generator ($8,000)

    Bars are the sum of the LOW end of each priced line item, multiplied by the starting quantity where a line is priced per unit, so they are a floor and not an estimate. 7 of 33 rows contribute nothing to them: four carry no figure at all (workload dependent, service dependent, hardware plus service, future node pricing), two are $0 software, and one is an order-of-magnitude entry rather than a number. All seven still appear in the table below. Nothing here is summed into a single figure for the whole build.

    Build sheet for the Mixpeek Appliance, by phase, with planning price ranges
    PhaseComponentPurposeExtends toPlanning range
    NOWStorage Pod 01: 4U 24-bayHot-swap server with ECC, BMC, HBA/backplane, redundant PSU; the literal chassis that later slides into the rack, bought now instead of a throwaway enclosureMakes durable storage independent from compute from day one; never gets replacedPopulate 24 bays over time; Storage-02/N; dense JBOD later$4,000-$8,000 before disks
    NOWSeagate Exos X24 24 TB SATA HDDEnterprise 3.5-inch SATA diskPopulate Storage-01 from day one; drives never migrate because the chassis never changes4 drives = 96 TB raw now; fill toward 24 bays over time, same chassis through V1$350-$500 each
    NOWZFS + S3-compatible object serviceDurable implementation + stable object APICanonical `s3://` namespace live from day one, not created during a later migrationBackend can become distributed object storage$0+ software
    NOWIndependent backupSeparate NAS/disks/cloud copyAvoid a single-copy eventGrows with corpusWorkload dependent
    V127-32U enclosed vertical rackFull-depth, lockable, wheeled, high-airflow cabinetClean movable mini data centerAdd pods until full; duplicate rack later$1,000-$2,500
    V1Compute Pod 01: 4U chassisFull-depth GPU chassis; server airflow/serviceability preferredReplaceable AI compute unitGPU #2 vertically; Compute-02/N horizontally$700-$2,000
    V1RTX PRO 6000 Blackwell 96 GB96 GB ECC NVIDIA accelerator; Server Edition preferred in qualified serverLarge coding/VLM/video inference, Mixpeek extraction, training/fine-tuningSecond 96 GB GPU; future RTX PRO/MGX/HGX/DGX nodes$13,000-$15,000
    V1Threadripper PRO 9975WX32C/64T, high PCIe capacityFFmpeg, Ray CPU workers, preprocessing, tokenization, DBs, GPU feedingAdd CPU/GPU nodes; higher-core node if measured need$4,000-$5,000
    V1ASUS Pro WS WRX90E-SAGE SEECC RDIMM, PCIe 5.0, dual 10GbE, AST2600 BMC/IPMIExpansion + RAM bandwidth + remote recoveryReserve x16 for GPU #2 and ConnectX NIC$1,200-$1,500
    V1512 GB ECC DDR5 RDIMMRegistered ECC memoryDataset staging, Ray, dataloaders, CPU offload, trainingScale toward ~2 TB; each node adds RAM$1,500-$3,000
    V12 × 2 TB mirrored boot NVMeRedundant system volumeUbuntu, configs, critical service stateLarger mirror/dedicated management storage later$300-$600
    V18 TB model/dataset NVMeFast reusable local tierWeights, HF cache, tokenized/hot datasetsAdd/larger enterprise NVMe per node$700-$1,500
    V14-8 TB high-endurance scratch NVMeSeparate high-write tierFrames/audio, checkpoints, optimizer state, temp tensorsAdd NVMe; shared hot data moves to Lustre at V2$500-$1,500
    V11600 W+ PSU / OEM redundant PSUGPU-capable powerStable compute power with headroomGPU #2 where envelope permits$500-$1,000
    V1Cooling/fansTR5/server cooling, front-to-back airflowSustained 24/7 operationOEM/liquid cooling for dense future accelerators$250-$700
    V1+ 8 × 20-24 TB enterprise SATA HDDsGrows the pool already running since NOW, to ~12 drives totalCanonical media, Mixpeek objects, datasets, artifacts~240-288 TB raw at ~12 drives; ~480-576 TB with all 24 bays full; then PB shelves$2,800-$4,000
    V1100 GbE ConnectX-class NICsHigh-speed node adaptersKeep storage traffic from starving GPUs100 → 200/400 GbE RDMA$500-$2,000/node
    V1100 GbE managed data switchInternal high-speed fabricCompute ↔ storage and future shared tierUpgrade to 200/400G; split fabrics later$1,500-$5,000
    V1Management switch/VLANSeparate OOB networkBMC/IPMI, UPS, switches, controllers, sensorsAdd all future nodes/racks$200-$750
    V1Kubernetes control nodeSmall dedicated x86 hostKeeps orchestration independent of GPU workers1 → 3 HA controllers$300-$800
    V1Local consoleSmall monitor + keyboard/mouse + shelfBreak-glass install/BIOS/network recoveryRack KVM later$150-$500
    V1Rack PDU + UPSMetered power + short ride-throughClean power and graceful shutdownLarger/redundant UPS as load grows$1,500-$4,000
    V1Rails/optics/DAC/SAS/power/sparesIntegration hardwareRepeatable, serviceable installationStandardize for future pods$500-$1,500
    V1Tailscale access planeSecure overlay for clients/adminMac/amux and engineers reach services without rack depending on MacAdd users/sites; substitute approved on-prem solution for true air-gapService dependent
    V2Lustre shared hot tierParallel POSIX filesystem over fast/RDMA fabricShared training/video working set; avoids node-local data dependencyAdd MDS/OSS/OST capacity$5,000-$25,000+
    V2Compute Pod 02Second GPU worker nodeHorizontal throughput + distributed trainingCompute-03/NFuture node pricing
    V2Kubernetes + GPU Operator + KubeRayCluster scheduling/distributed executionOne logical resource poolAdd workers horizontally$0+ software
    FUTURE200/400 GbE or InfiniBand compute fabricTightly coupled inter-node networkDistributed training / giant model parallelismFuture HGX/DGX/MGX nodes$10Ks+
    FUTUREDense Seagate-class JBOD/storage shelf60-106-drive expansionHundreds of TB → PB-scale object storageAdd shelves independently of compute$50K-$150K+
    OFF-GRIDStarlink + secure accessRemote WANCabin connectivityBackup WANHardware + service
    OFF-GRID48 V LiFePO4 bankLarge-format battery storageOvernight/low-solar operationAdd modules from measured kWh/day~$20K-$35K+ V1 scale
    OFF-GRIDPNW solar arrayGround array; planning target ~25-30 kW for year-round V1 resilienceGenerate rack energy + recharge batteriesExpand from measured load/site yield~$15K-$40K+ installed
    OFF-GRIDHybrid inverter + generatorAC conversion/charging + long-dark-period backupResilient off-grid operationParallel/larger equipment as compute grows$8K-$25K+

    Showing 33 of 33 components.

    Worked examples

    What the sheet above actually buys, in capacity terms. Every figure is arithmetic from the assumption printed beside it, so you can substitute your own bitrate or drive size and redo it. None of these are benchmarks: we publish prices and capacities, and we are not going to publish throughput numbers we have not measured.

    How much 1080p video fits in the V1 storage pod?

    Assumptions

    • ·12 x 24 TB enterprise SATA, the V1 starting population
    • ·Two 6-wide RAIDZ2 vdevs, so 4 of the 12 drives are parity
    • ·1080p H.264 at roughly 2 GB per hour
    Raw capacity12 x 24 TB = 288 TB
    After RAIDZ2 parity8 of 12 usable = 192 TB
    Usable after ~10% ZFS overhead and free-space headroom~173 TB
    At 2 GB per hour173,000 GB / 2 = ~86,000 hours

    Roughly 86,000 hours of 1080p video, about 9.9 years of continuous footage.

    Swap the bitrate for your own and the division is the only thing that changes. 4K at 8 GB per hour lands nearer 21,000 hours.

    What happens when you fill all 24 bays?

    Assumptions

    • ·The same pod, 24 bays populated at 24 TB
    • ·Four 6-wide RAIDZ2 vdevs, the same layout repeated
    Raw capacity24 x 24 TB = 576 TB
    After parity16 of 24 usable = 384 TB
    Usable after overhead~346 TB

    The pod roughly doubles without a second chassis, a second rack, or any change to the object namespace.

    This is the point of buying the 24-bay chassis at V1 rather than a 12-bay one. The expansion is drives, not architecture.

    What does Phase 0 buy before the rack exists?

    Assumptions

    • ·The final 24-bay storage server, bought now rather than a desktop enclosure
    • ·Populated with 4 x 24 TB enterprise SATA to start
    • ·Runs beside the desk on ordinary 1/10 GbE until the rack cabinet exists
    Raw capacity4 x 24 TB = 96 TB
    Filled to 8 bays8 x 24 TB = 192 TB
    Planning cost: chassis before disks$4,000 to $8,000
    Plus four drives at $350 to $500 each$1,400 to $2,000
    Purchase 0 target before backuproughly $5,400 to $10,000

    96 TB of usable corpus beside the desk now, on the exact chassis that later slides into the rack. There is no migration event, because nothing moves.

    Mirrored or parity layouts cut the usable figure; this is the raw number. It costs more than the desktop-enclosure plan it replaced, and what the extra buys is that nothing here is ever replaced and the canonical s3:// namespace exists from day one instead of being created during a later migration.

    Software stack

    All of it runs in the rack. Nothing in this list depends on a hosted service to function.

    • Mixpeek local stack
    • Ray / KubeRay
    • Kubernetes
    • NVIDIA GPU Operator
    • vLLM
    • SGLang
    • TensorRT-LLM
    • NeMo
    • Megatron
    • PyTorch
    • Hugging Face
    • Qdrant
    • S3-compatible object service
    • ZFS
    • Lustre (V2)
    • Prometheus / Grafana-class observability

    The one rule behind the whole sheet

    Every NOW-phase line is chosen so it is not thrown away at V1, and the plan is stricter than that now: the storage server bought in Phase 0 is the same physical unit that later slides into the rack, so there is no migration event to survive. It costs more up front than a desktop enclosure would, and what that buys is that the canonical s3:// namespace exists from the first day rather than being created during a later copy-and-verify. That constraint is why the build order starts with storage: it is the only layer where buying early does not mean buying twice.

    The purchase sequence and scaling model · Common questions

    Not sure what you would actually need?

    Tell us the corpus, the workloads and the constraints, and we will size it against the build sheet: drive count, GPU class, fabric, power, and which phase to start at. You own the hardware either way, and the plan is published in full whether you build it yourself or have us do it.

    It is fully S3-compatible, with no code change. The object endpoint speaks the S3 API, so anything already talking to S3 points at a new endpoint URL and keeps working. Same SDKs, same bucket and key layout, same tooling.