Netris Announces $15M From a16z to Automate AI Data Centre Networking for Neoclouds
Netris secured $15 million in Series A funding led by Andreessen Horowitz to accelerate growth in AI infrastructure automation. The company’s platform automates data center networking setup and operations, significantly reducing the time GPU clusters remain idle before generating revenue. Netris distinguishes itself by running networking automation entirely in hardware rather than via software-defined networking, ensuring it can handle the high traffic volumes of AI workloads. The solution is ve
Analysis
TL;DR
- Netris secured $15 million in Series A funding led by Andreessen Horowitz to accelerate growth in AI infrastructure automation.
- The company’s platform automates data center networking setup and operations, significantly reducing the time GPU clusters remain idle before generating revenue.
- Netris distinguishes itself by running networking automation entirely in hardware rather than via software-defined networking, ensuring it can handle the high traffic volumes of AI workloads.
- The solution is vendor-agnostic, supporting both Nvidia and AMD environments, and is currently deployed across over 35 global GPU clusters totaling approximately one million GPUs.
- The platform relies on deterministic algorithms rather than AI for its own operations to ensure the precision and repeatability required for configuring thousands of switches.
Why It Matters
This development highlights a critical bottleneck in the current AI boom: the operational complexity and time lag in provisioning networking infrastructure for massive GPU clusters. By addressing the gap between hardware availability and operational readiness, Netris enables neocloud operators to monetize expensive assets faster, directly impacting the ROI of large-scale AI deployments. For the industry, it signals a shift toward specialized, hardware-accelerated networking solutions that prioritize determinism and speed over general-purpose software abstractions.
Technical Details
- Hardware-Accelerated Automation: Unlike traditional software-defined networking (SDN), Netris executes automation logic directly in hardware, allowing it to manage traffic volumes and switch configurations at speeds that software-based solutions cannot match.
- Deterministic Algorithms: The platform avoids using AI for its own control plane, relying instead on deterministic algorithms refined over eight years to guarantee precision, stability, and repeatability in complex network setups.
- Multi-Tenancy and Abstraction: The system provides hardware-level abstraction and multi-tenancy isolation, enabling operators to securely serve multiple customers from shared infrastructure without performance degradation.
- Vendor Agnosticism: The software is compatible with diverse hardware ecosystems, specifically supporting both Nvidia and AMD server environments, reducing lock-in risks for operators.
- Scale and Deployment: The technology is live across more than 35 GPU clusters globally, managing approximately one million GPUs for major operators such as Lightning AI, Foxconn, HPE, TensorWave, and Telus.
Industry Insight
- Operational Efficiency as a Competitive Advantage: As AI infrastructure costs soar, the ability to rapidly provision and configure networking becomes a key differentiator. Companies that reduce idle time for GPU clusters will gain significant cost advantages and faster time-to-market.
- Rise of Neocloud Operators: The success of Netris underscores the growing importance of neocloud providers who offer specialized, automated infrastructure. Traditional data center management tools are insufficient for the scale and speed of modern AI training and inference workloads.
- Preference for Determinism in Critical Infrastructure: The rejection of AI-driven automation for core networking tasks suggests that for mission-critical, high-stakes infrastructure, deterministic and predictable systems will remain preferred over probabilistic AI models, at least in the near term.
Disclaimer: The above content is generated by AI and is for reference only.