Tenstorrent’s Galaxy Blackhole AI servers escape the event horizon

2026-04-28 · Source: The Register: Enterprise Technology News and Analysis · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Cloud Computing & IT Infrastructure · Depth: Intermediate, short

Summary

Tenstorrent has announced the general availability of its Galaxy Blackhole AI compute platform, a RISC-V-based system featuring 32 Blackhole accelerators within a 6U chassis, priced at $110,000. Each Galaxy system offers 1 TB of GDDR6, 16 TB/s memory bandwidth, and 23 petaFLOPS of dense FP8 performance. The accelerators are interconnected via a 100 Tbps Ethernet mesh, which can be extended to form larger clusters, such as the $440,000 base Galaxy Supercluster with four systems. Tenstorrent claims significant software stack improvements, enabling a four-node Supercluster to process a 100,000-token DeepSeek V3 prompt in under four seconds and generate 720p video faster than real-time. The company also states that "Ninety percent of models from Hugging Face just run on Tenstorrent."

Key takeaway

For CTOs and VP of Engineering evaluating AI infrastructure, Tenstorrent's Galaxy Blackhole platform presents a compelling alternative to higher-priced solutions like Nvidia's DGX. Its scalable mesh architecture and claimed performance for large language models and video generation, coupled with a significantly lower entry cost, warrant consideration for your next-generation AI deployments. Investigate its real-world performance with your specific workloads and model types.

Key insights

Tenstorrent's Galaxy Blackhole offers a scalable, cost-effective AI compute platform with competitive performance for large models.

Principles

• RISC-V architecture enables specialized AI acceleration.
• Dense mesh networks enhance multi-node AI system scalability.

Method

Tenstorrent uses a Python-based programming interface for writing optimized kernels, facilitating the porting and performance tuning of new AI models on its hardware.

In practice

• Deploy Galaxy Blackhole for large language model inference.
• Utilize for high-throughput video generation tasks.
• Explore for cost-efficient AI infrastructure scaling.

Topics

• Tenstorrent Galaxy Blackhole
• Blackhole Accelerators
• RISC-V Architecture
• AI Server Systems
• Ethernet Mesh Interconnect

Best for: CTO, VP of Engineering/Data, AI Engineer, AI Architect, MLOps Engineer, Director of AI/ML

Related on AIssential

• Tenstorrent Previews Large Compute Cluster, Generates Video Faster Than Real Time — Tenstorrent's new hardware accelerates video generation by 10x, breaking the real-time barrier for five-second videos.
• NVIDIA Partners With Microsoft on Unified Stack for Agentic AI Deployment, From Windows Devices to Cloud to Local — NVIDIA and Microsoft are building a full-stack, secure, and scalable ecosystem for agentic AI, from edge devices to cloud.
• XAI's Radical Plan: Data Centers In Space — XAI plans orbital AI data centers and advanced automation, leveraging SpaceX's capabilities to overcome terrestrial compute limitations.
• Qualcomm in Talks to Buy Tenstorrent to Expand AI Chip Capabilities — Qualcomm's potential acquisition of Tenstorrent signals strategic expansion in the competitive AI chip market.
• Google's TPU AI Compute Arbitrage — Google Cloud uses custom TPUs to create a structural cost advantage, funding competitive GPU offerings.
• SiFive $400M Round Highlights New CPU Battleground for Agentic AI Demand — RISC-V is gaining significant investment and traction as a customizable, power-efficient CPU alternative for agentic AI in data centers.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by The Register: Enterprise Technology News and Analysis.