Hala Nexus (STHT1)
Revolutionizing Business with AI Desktops
AI Desktop Workstation powered by AMD Ryzen AI Strix Halo. Run massive LLMs locally with up to 235B parameters. Unprecedented on-device AI performance.
Powerhouse Features
Everything for personal workstations to team deployments.
80 TOPS NPU
80 TOPS NPU dedicated AI processing. Run any LLM locally with zero latency.
RDNA 3.5 Integrated Graphics
AMD integrated graphics with up to 96GB shared memory. Discrete-level performance for AI and creative workflows.
Dual M.2 NVMe Storage
2× M.2 2280 PCIe×4 SSD. SD4.0 card reader for SDXC. Expand for datasets and model caching.
Industrial Connectivity
WiFi 7 (802.11be), Bluetooth 5.0+, 2.5G Ethernet. Deploy anywhere, connect reliably.
A New Class Of AI
Desktop
H13 combines high-power CPU cores, RDNA 3.5 graphics, and an industry-leading NPU into a single compact system. It is designed for developers, teams, and enterprises that need local, scalable, high-performance AI compute without relying on cloud infrastructure.
🚀Unprecedented AI Acceleration
- 80 TOPS NPU — industry leading
- Optimized for AI inference, Copilot+ PC, and LLM workloads
- High power range (120W / 132W) for sustained performance
🖥Discrete-Level Graphics Performance
- Built-in RDNA 3.5 graphics (up to 40CU)
- Up to 2.5× graphics performance vs previous gen AI chips
- Competitive with RTX 4070 mobile performance
- Superior thermal design for sustained local compute
🧠Massive Unified Memory
- Up to 128GB LPDDR5 (8533MT/s)
- Allocate up to 96GB as variable graphics memory
- Ideal for large AI models and high-resolution workloads
Hala Nexus vs. The Giants
Why settle for cloud-taxed, proprietary hardware when you can own your data and your compute? Hala delivers industrial-grade local AI at a fraction of the workstation cost.
| Feature | Hala Nexus | Mac Studio (M2 Ultra) | Dell Precision 3280 |
|---|---|---|---|
| AI Performance (NPU) | 80 TOPS | 15.8 TOPS | 11 TOPS |
| Unified Memory | 128GB (LPDDR5x) | 128GB (Unified) | 128GB (Split VRAM) |
| Memory Bandwidth | 256 GB/s | 800 GB/s | 64 GB/s |
| Local LLM Cap | Up to 235B Params | High (Proprietary) | Limited (20GB VRAM) |
| Privacy / Stack | 100% Zero-Cloud | Apple Intelligence | Windows / Cloud-First |
| Starting Price | $2,899 | $3,999+ | $3,500+ |
Endless Possibilities With The Ultras Powered H13
Edge AI Inference
Run AI models locally for real-time processing without cloud dependency
Local LLM Deployment
Deploy large language models on-premises for privacy and performance
AI-Powered Search
Search and retrieve information instantly with semantic capabilities
3D Rendering & Content
High-performance graphics processing for visualization and creation
Built For All, Built
For Everyone
From personal to enterprise, one system for all, RDNA 3.5 graphics, and an industry-leading NPU into a single compact system. Designed for developers, teams, and enterprises that need local, scalable, high-performance AI compute without relying on cloud infrastructure.
AI Research
Experiment with large models, fine-tune locally, no vendor dependency.
- LLM Training
- Model Research
- Local Deployment
Creator Studio
Video editing, 3D rendering, and AI-assisted production in one box.
- Video Editing
- 3D Rendering
- VFX & AI
Gaming & Content
Local LLM AI, 2.5X graphics performance, and real-time content generation.
- AAA Gaming
- AI-Generated Content
- 3D Modeling
Enterprise
Scalable clusters for regulated industries and high-security environments.
- Data Sovereign
- Multi-Unit Scale
- Zero Trust
Key Specifications
See what innovators are building
AI Acceleration
80 TOPS
Dedicated 80 TOPS NPU engine for edge AI inference. Run 235B parameter models locally at high speeds.
Model Compatibility: Qwen3, Llama, DeepSeek, and any OSS LLM
Memory
Up to 128GB
LPDDR5 @ 8533MT/s
Graphics
RDNA 3.5
40CU Flagship Performance
Storage
2× M.2 2280
PCIe Gen 4×4 NVMe SSD
Processor
AMD Strix Halo
AI 395 · 16-core
Networking
1× 2.5G LAN
WiFi 7 · Bluetooth 5+ · USB4
Power
240W
20V/12A external adapter
Ready To Experience AI Dominance?
Check More Products
Discover other amazing products to help drive local AI efficiencies in your organization.
Hala Pulse(STHT1)
Coming Soon
Ultra-compact 127mm AI workstation with 16-core Strix Halo AI 395 and 80 TOPS NPU.
Hala Apex(STHT1)
Coming Soon
12-core Strix Halo AI 385 workstation with 80 TOPS NPU and high-performance RDNA 3.5 graphics.
Hala Core(STHT1)
Coming Soon
Entry-level AI node with 8-core Ryzen 7840, 16 TOPS NPU, and 32GB RAM.
Motherboard STH-1
Coming Soon
Thin Mini-ITX Strix Halo motherboard with 64GB LPDDR5X, 1× PCIe 4.0 x4 + 1× PCIe 3.0 x4, and dual HDMI 2.1.
Motherboard STH-T1-8
Coming Soon
8-core Strix Halo motherboard configuration with 128GB LPDDR5X, dual M.2 PCIe 4.0 x4, and 19V DC-in.