Netris secures $15 million in Series A funding from a16z to streamline the networking processes that hinder GPU cloud performance.
Netris secured $15 million from Andreessen Horowitz for its GPU network automation platform, which is currently operational in over 35 clusters with an impressive 800 percent annual revenue growth. The Santa Clara-based startup focuses on automating the networking layer within GPU data centers, and this funding round follows its rapid growth, alongside deployments at various GPU clusters globally, including those operated by Lightning AI, Visionbay (backed by Foxconn), Hewlett Packard Enterprise, TensorWave, and Telus.
Netris addresses a less-discussed challenge in the AI infrastructure surge. While significant investments have been made in acquiring GPUs and establishing data centers, the critical step of configuring the network connecting thousands of servers can take months, often causing costly delays. According to the company, a single GPU server has at least three north-south connections, 16 east-west connections, and four NVL72 links. Each time a tenant is added, resized, or removed, the network must undergo simultaneous reconfiguration across multiple layers, often involving hundreds or thousands of switches, where a single error can disrupt the cluster or compromise customer data.
Netris offers software for network switches and a platform that automates the setup, configuration, and management for neocloud operators. This platform also enables network abstraction, allowing hardware configurations to be modified as needed, while isolating servers and resources at the hardware level to facilitate multi-tenancy. The company refers to this strategy as NAAM, short for Network Automation, Abstraction, and Multi-Tenancy.
CEO Alex Saroyan explained to TechCrunch that traditional software-defined networking is inadequate for AI workloads due to the high traffic volume, necessitating a hardware-accelerated approach. He noted that Netris has spent eight years developing hardware-accelerated network automation.
Nvidia, which has invested over $40 billion in AI infrastructure this year, recognized Netris's potential early on and began recommending it to customers after being impressed by a demonstration two years ago. Saroyan confirmed that the platform is vendor-agnostic and compatible with networking equipment from both Nvidia and AMD.
The startup claims that its platform is currently operational in over 35 GPU clusters that collectively house approximately one million GPUs. The neocloud sector has recently experienced an influx of funding, with companies like Runpod reaching billion-dollar valuations due to the AI compute deficiency. Netris, however, functions within a different layer of the stack, providing necessary infrastructure software to GPU cloud operators rather than competing for compute customers.
Guido Appenzeller, a partner at a16z and former co-founder of Big Switch Networks who served as CTO at VMware's Cloud and Networking division, led the funding round and will join Netris's board. He stated that "GPU clusters run across many fabrics at once, and legacy automation was never designed for that," asserting that Netris is becoming the standard platform for AI cloud operators.
Interestingly, Netris does not incorporate AI into its own product; instead, it utilizes deterministic algorithms developed prior to the AI wave. Saroyan emphasized that creativity is not a desirable quality for a system managing thousands of switch configurations, arguing that network configuration requires persistence and repeatability rather than creativity.
Founded in 2018, Netris operates teams across the U.S., U.K., Taiwan, Australia, Armenia, and India, with plans to launch in Singapore this year. The company intends to use the new funding to hire more engineers and sales personnel, broaden support for additional hardware vendors, and enhance the capabilities of its automation platform.
Other articles
Klue reports that the hackers responsible for stealing customer data are removing it, but another group has surfaced with demands for extortion.
Klue reports that the hacking group Icarus is removing stolen customer data; however, a second unnamed group asserts that it possesses the data and is directly extorting the victims.
Microsoft increases Xbox prices for the third time in 13 months, with the Series X now costing $800 due to memory shortages.
Starting August 1, Xbox consoles will see a price increase of $100 to $150, bringing the cost of the Series X to $800. Microsoft attributes this rise to the more than doubling of memory expenses.
Taktile secures $110 million to entrust AI with banking decisions.
Taktile has secured $110 million in a Series C funding round led by Goldman Sachs, enabling banks and insurers to automate their most critical decisions using AI agents.
Coval secures $28 million to evaluate the performance of AI voice agents under pressure.
Coval has secured $28 million in funding, led by Norwest, to replicate and evaluate enterprise AI voice agents prior to their performance in actual customer calls.
Klue reports that the hackers responsible for stealing customer data are in the process of deleting it, while a second group has come forward with demands for extortion.
Klue reports that the hacking group Icarus is removing stolen customer information, while a second unnamed group asserts that they possess the data and are directly extorting the victims.
Google Finance has officially left beta, introducing a dedicated Android app, portfolio tracking features, and AI-driven scheduled briefings.
Google Finance exits beta with the launch of a new Android app, portfolio dashboards, an AI research tool, and scheduled market briefings powered by Gemini.
Netris secures $15 million in Series A funding from a16z to streamline the networking processes that hinder GPU cloud performance.
Netris has secured $15 million in funding from Andreessen Horowitz to expand its network automation platform, which is currently operational in over 35 GPU clusters globally.