The hardware, one page at a time.
Each rig has specs, live-ish pricing, our verdict, and a one-click path into the arena.
By category
By vendor
By spec — narrow to what fits the PO
Sort by
- 1X Neo
1X Technologies · edge
MSRP · $20,000
Soft-tendon home humanoid: tendon-driven actuation, US preorder open, deliveries start 2026.
2026-05-23 · verified 3w ago
- Agility Robotics Digit
Agility Robotics · edge
Warehouse-deployment humanoid: logistics-first form factor, dual-CPU compute, commercial pilot at Amazon and GXO.
2026-05-23 · verified 3w ago
- Apptronik Apollo
Apptronik · edge
NASA-Valkyrie-derived industrial humanoid: hot-swap battery, force-control architecture, Mercedes-Benz pilot.
2026-05-23 · verified 3w ago
- Boston Dynamics Atlas
Boston Dynamics · edge
All-electric humanoid: 56 DoF, IP67, 50 kg instant payload, Google DeepMind brain, NVIDIA sim, Hyundai factories.
2026-05-23 · verified 3w ago
- Boston Dynamics Spot
Boston Dynamics · edge
MSRP · $74,500
Commercial quadruped inspection platform: IP54, 90-minute runtime, 14 kg payload, the canonical robot dog of industrial deployment.
2026-05-23 · verified 3w ago
- Figure 02
Figure · edge
OpenAI-partnered humanoid with 16-DoF hands, 6 RGB cameras, 2.25 kWh torso pack. Pilot inside BMW Spartanburg final assembly.
2026-05-23 · verified 3w ago
- Sanctuary AI Phoenix
Sanctuary AI · edge
Sixth-generation humanoid from Vancouver: 20-DOF hands, the Carbon AI control system, bipedal general-purpose form factor.
2026-05-23 · verified 3w ago
- Tesla Optimus
Tesla · edge
Vision-only humanoid built on Tesla's FSD neural-net stack. Gen 2 shipped 2023; Gen 3 production tooling targeted for summer 2026.
2026-05-23 · verified 3w ago
- Unitree B2
Unitree Robotics · edge
MSRP · $100,000
Industrial quadruped at 60 kg with 40 kg walking payload, IP67 sealing, 5+ hour endurance, and Jetson Orin NX compute optionality.
2026-05-23 · verified 3w ago
- Unitree G1
Unitree Robotics · edge
MSRP · $13,500
Open-store humanoid: 23-DoF base / 43-DoF EDU, sub-$14k starting price, ships from Unitree's web shop today.
2026-05-23 · verified 3w ago
- Unitree H1
Unitree · edge
MSRP · $90,000
Open humanoid: 23 DoF, 3.3 m/s walking record, swappable 864 Wh pack, optional Jetson Orin NX. Commercial — sub-$90k list.
2026-05-23 · verified 3w ago
- AMD Instinct MI300X
AMD · accelerator
CDNA 3 OAM accelerator with HBM3 and TSMC 5nm/6nm chiplets.
2026-05-01 · verified 1mo ago
- AMD Instinct MI325X
AMD · accelerator
CDNA 3 refresh: HBM3E in the same OAM envelope as MI300X.
2026-05-01 · verified 1mo ago
- AMD Instinct MI355X
AMD · accelerator
CDNA 4 OAM: HBM3E, native FP4/FP6, liquid-cooled flagship.
2026-05-01 · verified 1mo ago
- Apple Mac Pro (M2 Ultra, 2023)
Apple · mac
MSRP · $6,999
Tower workstation with M2 Ultra: large unified memory at 800 GB/s. Discontinued March 2026.
2026-05-01 · verified 1mo ago
- Apple MacBook Pro 16-inch, M4 Max (2024)
Apple · mac
MSRP · $3,499
Mobile workstation with M4 Max: large unified memory in a 16.2" Liquid Retina XDR chassis.
2026-05-01 · verified 1mo ago
- Aurora
Intel / HPE · workstation
Argonne's Intel-powered exascale system — Xeon Max CPUs and Intel Data Center GPU Max accelerators.
2026-05-01 · verified 1mo ago
- AWS Trainium2
AWS · cloud-model
Amazon's training/inference silicon: HBM3e per chip; UltraServers via NeuronLink scale to dozens of chips.
2026-05-01 · verified 1mo ago
- Cerebras Wafer-Scale Engine 3
Cerebras · accelerator
Wafer-scale AI engine: trillions of transistors and hundreds of thousands of cores on a single TSMC 5nm die.
2026-05-01 · verified 1mo ago
- El Capitan
HPE / AMD · workstation
LLNL's exascale top-of-list — AMD MI300A APUs across millions of cores.
2026-05-01 · verified 1mo ago
- Frontier (OLCF-5)
HPE / AMD · workstation
Oak Ridge's exascale system — first to break the ExaFLOPS barrier on HPL; AMD MI250X plus EPYC.
2026-05-01 · verified 1mo ago
- Google Cloud TPU v5p
Google · cloud-model
Google's largest training TPU pod — 3D-torus interconnect, BF16-strong chips, accessible as Cloud TPU.
2026-05-01 · verified 1mo ago
- Groq Language Processing Unit
Groq · accelerator
Inference-only ASIC trading HBM for on-die SRAM — deterministic latency for LLM serving.
2026-05-01 · verified 1mo ago
- Intel Gaudi 3 HL-325L
Intel · accelerator
OAM accelerator with HBM2E and on-die RoCE Ethernet for scale-out without external NICs.
2026-05-01 · verified 1mo ago
- NVIDIA B100
NVIDIA · accelerator
Blackwell SXM module designed as a drop-in for HGX H100 baseboards. Lower-TDP sibling of B200.
2026-05-01 · verified 1mo ago
- NVIDIA B200
NVIDIA · accelerator
Blackwell SXM module. Dual-die GPU with HBM3e, fifth-gen NVLink. Building block of HGX B200, DGX B200, and GB200.
2026-05-01 · verified 1mo ago
- NVIDIA DGX B200
NVIDIA · workstation
Air-cooled DGX. Eight Blackwell GPUs on an HGX baseboard, dual Xeon Platinum, BlueField-3, ConnectX-7.
2026-05-01 · verified 1mo ago
- NVIDIA DGX H200
NVIDIA · workstation
Eight-way H200 server with HBM3e, FP8 Tensor Cores, dual Xeon Platinum, NVSwitch fabric.
2026-05-01 · verified 1mo ago
- NVIDIA DGX Spark
NVIDIA · workstation
MSRP · $4,699
GB10 Grace Blackwell mini-workstation: unified memory, ConnectX-7, desk-sized chassis.
2026-05-01 · verified 1mo ago
- NVIDIA Eos DGX SuperPOD
NVIDIA · workstation
NVIDIA's in-house Hopper supercomputer — exa-class FP8 across hundreds of DGX H100 nodes.
2026-05-01 · verified 1mo ago
- NVIDIA GB200
NVIDIA · accelerator
One Grace CPU coherently linked to two Blackwell GPUs over NVLink-C2C. Unit cell of the NVL72.
2026-05-01 · verified 1mo ago
- NVIDIA GB200 NVL72
NVIDIA · workstation
Liquid-cooled rack — Blackwell GPUs and Grace CPUs as one NVLink domain, exaflop-class FP4.
2026-05-01 · verified 1mo ago
- NVIDIA GeForce RTX 5070 Ti Founders Edition
NVIDIA · gpu
MSRP · $749
Blackwell mid-tier: 16 GB GDDR7 on a 256-bit bus, 8,960 CUDA cores, 300 W TGP.
2026-05-01 · verified 1mo ago
- NVIDIA GeForce RTX 5080 Founders Edition
NVIDIA · gpu
MSRP · $999
Blackwell second-tier: 16 GB GDDR7 at 30 Gbps, 10,752 CUDA cores, 360 W TGP, 2-slot reference cooler.
2026-05-01 · verified 1mo ago
- NVIDIA GeForce RTX 5090 Founders Edition
NVIDIA · gpu
MSRP · $1,999
Blackwell flagship: 32 GB GDDR7, 21,760 CUDA cores, 575 W TGP, 2-slot Founders form factor.
2026-05-01 · verified 1mo ago
- NVIDIA GH200 Grace Hopper
NVIDIA · accelerator
Grace Arm CPU and Hopper GPU on one module, joined by NVLink-C2C, with HBM3e on the GPU side.
2026-05-01 · verified 1mo ago
- NVIDIA H100
NVIDIA · accelerator
Hopper-generation SXM accelerator: HBM3, fourth-gen Tensor Cores, the workhorse of the LLM training era.
2026-05-01 · verified 1mo ago
- NVIDIA H200
NVIDIA · accelerator
Hopper-generation SXM accelerator with HBM3e, drop-in for HGX H100 baseboards.
2026-05-01 · verified 1mo ago
- NVIDIA HGX B200
NVIDIA · accelerator
Eight-GPU Blackwell baseboard. HBM3e, NVLink 5 fabric, foundation of OEM Blackwell servers.
2026-05-01 · verified 1mo ago
- NVIDIA Jetson AGX Thor
NVIDIA · edge
MSRP · $3,499
Blackwell-generation edge module: native FP4, LPDDR5X, Arm Neoverse-V3AE, configurable power envelope.
2026-05-01 · verified 3w ago
- NVIDIA RTX PRO 4500 Blackwell
NVIDIA · workstation
Workstation entry: 32 GB GDDR7 ECC, 10,496 CUDA cores, 200 W, dual-slot.
2026-05-01 · verified 1mo ago
- NVIDIA RTX PRO 5000 Blackwell
NVIDIA · workstation
Workstation tier: 48 GB or 72 GB GDDR7 ECC, 14,080 CUDA cores, 300 W, dual-slot.
2026-05-01 · verified 1mo ago
- NVIDIA RTX PRO 6000 Blackwell
NVIDIA · workstation
Workstation flagship: 96 GB GDDR7 ECC, 24,064 CUDA cores, 600 W max, dual-slot.
2026-05-01 · verified 1mo ago
- SambaNova SN40L Reconfigurable Dataflow Unit
SambaNova · accelerator
Three-tier-memory dataflow accelerator: on-chip SRAM, HBM3, and TB-class DDR5 per socket.
2026-05-01 · verified 1mo ago
- The Stargate Project
OpenAI / Oracle / SoftBank / MGX · workstation
Multi-site AI buildout — gigawatts of planned capacity, millions of chips, primary site Abilene, Texas.
2026-05-01 · verified 1mo ago
- Alibaba Qwen3-72B-Instruct (Q4 quant)
Alibaba · cloud-model
Frontier-adjacent. Open weights. Your GPU, your rules.
2026-04-24 · verified 1mo ago
- Anthropic Claude Opus 4.7
Anthropic · cloud-model
The reference Soul in the canonical boss rig.
2026-04-24 · verified 1mo ago
- Apple Mac Studio (M3 Ultra)
Apple · mac
MSRP · $7,999
Up to 512 GB unified memory. Patient, capacious, cool.
2026-04-24 · verified 1mo ago
- MSI GeForce RTX 5090 Suprim Liquid
MSI · gpu
MSRP · $2,499
Factory-liquid 5090. 32 GB. No throttle.
2026-04-24 · verified 1mo ago
- NVIDIA GeForce RTX 4090 Founders Edition
NVIDIA · gpu
MSRP · $1,599
24 GB. Still fast. Still honest.
2026-04-24 · verified 1mo ago
- OpenAI GPT-5 Core
OpenAI · cloud-model
Broadest tool library. Fastest structured output.
2026-04-24 · verified 1mo ago
