← All hardware
$1,399 PCIe AI dev card with 32GB and open software
Pros
- Very accessible price
- Open, hackable software
- 32GB memory for the price
- Great for AI hardware dev/learning
Cons
- Recent firmware tensor-core downgrade (120 vs 140 cores)
- Smaller ecosystem than CUDA
- More effort to deploy
- Not for plug-and-play frameworks
✓ Where it shines / best for
- Developers and researchers wanting an open-source alternative to CUDA for AI inference/training
- Workstation-class AI model development and experimentation
- Building small multi-card AI clusters via native Ethernet links
✕ Not the best fit for
- Plug-and-play users expecting a mature CUDA-equivalent ecosystem
- Gaming or graphics rendering workloads
- Hyperscale production deployments needing turnkey vendor support
Features
- ✓ Inference
- ✓ Training
- ✓ Open source
- ✓ AI Accelerator
- ✓ Open Source Stack
- ✓ Risc V
- ✓ Ethernet Scaleout
- ✓ GDDR6
- ✓ PCIe Gen5
Pricing
| Plan | Price | Billing | Notes |
|---|---|---|---|
| Blackhole p150a | $1,299 | one-time | Active-cooled single-card variant; PCIe Gen5 |
| Blackhole p150b | $1,399 | one-time | Variant with 4x QSFP-DD Ethernet ports for multi-card scaling |
| Blackhole p100a | $999 | one-time | Lower-tier sibling card in same family for reference |
Pricing verified from the official source. Prices change often — confirm on the vendor's site before buying.
Specifications
| use | Developer/workstation AI inference |
| power | ~300W class |
| memory | 32GB |
| performance | ~664 TFLOPs |
| architecture | Blackhole (RISC-V Tensix) |
Sponsored
A full review is being generated for this product and will appear here shortly.