In short
Netris has raised a $15 million Series A led by Andreessen Horowitz to expand its network automation software for AI neocloud operators. The startup says its platform can help GPU clusters go live faster and manage multi-tenant infrastructure more efficiently.
- Netris raised $15 million in a Series A round led by Andreessen Horowitz.
- The startup automates network setup and operations for AI GPU clusters and neoclouds.
- Netris says it is deployed across 35+ GPU clusters worldwide, covering about 1 million GPUs.
- Nvidia previously recommended the company to customers after an early demo.
- The funding will support hiring, sales expansion and broader hardware support.
The rush to build AI infrastructure has created a new bottleneck: not the GPUs themselves, but the time it takes to turn a warehouse full of networking gear into a working cloud. Netris, a startup focused on automating that process, has raised $15 million in a Series A round led by Andreessen Horowitz as demand rises from so-called neocloud operators racing to bring AI capacity online faster.
The company says its software helps data center operators configure and manage network switches, isolate customers at the hardware layer, and make changes across large GPU clusters without the slow, manual work that can delay launches for months. In an industry where expensive accelerators can sit idle while infrastructure teams finish setup, that promise has become increasingly valuable.
Netris is already deployed in more than 35 GPU clusters worldwide, covering roughly 1 million GPUs, according to the company. Its customers include Lightning AI, Foxconn, Visionbay, Hewlett Packard Enterprise, Tensorwave and Telus, among others. Nvidia has also played an important role in the startup’s momentum after seeing an early demo of the technology and recommending Netris to customers.
The infrastructure race behind the AI boom
Artificial intelligence has not only transformed software development and model training; it has also sparked a scramble to build the physical infrastructure needed to support it. A wave of new providers has emerged to serve AI workloads with specialized data centers, often marketed as neoclouds. These companies are trying to stand up GPU-rich facilities quickly enough to meet demand from model builders, startups and enterprises.
That business model sounds straightforward: buy the chips, install the networking equipment, connect storage, and start selling compute. In practice, the process is far more complicated. Operators must ensure that thousands of devices are correctly configured, securely segmented and ready to serve multiple customers without errors or downtime.
For large cloud giants, that challenge has long been part of the playbook. Companies such as Microsoft, Google, Amazon, Oracle, Digital Realty, Equinix and NTT have spent years building custom systems and large engineering organizations to manage network operations at scale. Smaller neocloud businesses, by contrast, often lack the people and the tooling to automate those tasks efficiently.
That gap is where Netris sees its opportunity.
What Netris actually does
Netris describes its platform as a way to automate the configuration and operation of data center networks, with a particular focus on GPU clusters used for AI training and inference. The startup’s software runs on network switches and also connects to switch infrastructure to help operators bring facilities live more quickly.
The core idea is to reduce the amount of manual work required to provision and maintain a cluster. Instead of engineers making repetitive changes by hand across thousands of links and devices, Netris aims to push those changes through a system that is persistent, repeatable and designed for large-scale operations.
The company also emphasizes two other capabilities that matter to neocloud operators:
- Network abstraction, which allows hardware configurations to be changed without forcing operators to rebuild their network management approach from scratch.
- Multi-tenancy isolation, which separates customers and resources at the hardware layer so multiple clients can safely share the same infrastructure.
That combination matters because AI clusters are not generic server farms. They are high-throughput environments where networking performance is critical, traffic patterns are intense and customers expect flexible access to compute resources without compromising security or stability.
Why hardware acceleration matters
Netris argues that traditional software-defined networking tools are not enough for this environment. While SDN has been a standard way to manage modern data centers, the company says AI workloads create traffic levels that demand a more hardware-centered approach.
Alex Saroyan, Netris’s chief executive, said GPU cluster operators need to make configuration changes constantly and that software-only approaches are too slow for the scale and speed of AI infrastructure. He argued that the company’s platform offers a hardware-accelerated alternative built for repetitive, high-volume network changes rather than creative decision-making.
In Saroyan’s view, AI infrastructure requires precision more than spontaneity. The problem is not deciding what to do next; it is executing the same network change reliably across thousands of switch ports and links. That distinction helps explain why Netris says it has spent years building automation tools before the current AI boom fully arrived.
A startup that began before the AI hype
One detail that sets Netris apart from many newer AI infrastructure startups is that its technology predates the current generative AI surge. Saroyan said the company had been working on its automation algorithms for years before AI demand exploded, positioning Netris to benefit when GPU cluster operators began facing more severe networking challenges.
The company also says it does not rely on AI inside its own product. Rather than using machine learning to make decisions about network operations, Netris uses algorithms designed earlier for deterministic automation tasks. That choice reflects a practical philosophy: if the job is to push thousands of consistent configuration updates, the software should behave predictably every time.
This approach also gives Netris a different pitch from the many startups that wrap AI branding around infrastructure tools. Its claim is not that AI runs the platform, but that AI-driven data centers create a problem only disciplined automation can solve.
Nvidia’s early interest helped validate the idea
Nvidia’s involvement has added credibility to Netris’s story. The chipmaker reportedly saw a demo of the company’s technology about two years ago and was impressed enough to recommend it to customers. That kind of endorsement can carry unusual weight in a market where GPU supply, network performance and deployment speed are tightly linked.
For neocloud operators, support from a key hardware ecosystem player can make a meaningful difference. It can shorten the sales cycle, build trust and help customers view a startup as part of the broader AI infrastructure stack rather than as an untested point solution.
According to Netris, the company is now deployed in more than 35 GPU clusters globally, with about 1 million GPUs under management across those environments. The customer list spans AI-native firms, industrial companies and telecom operators, suggesting the market for automation in GPU networking is broader than just emerging AI labs.
Why Andreessen Horowitz invested now
With the new funding, Netris is moving from technical validation to scaling its business. Andreessen Horowitz led the Series A round, and partner Guido Appenzeller is joining the startup’s board. The investment suggests a16z sees a category forming around infrastructure software that helps neoclouds launch faster and operate more efficiently.
The timing is notable. As more companies enter the AI infrastructure market, the race is increasingly about time to revenue. Every month spent configuring networks and solving deployment problems is a month of foregone compute sales, especially when expensive GPUs are already installed and depreciating.
For investors, that creates a straightforward thesis: if neoclouds are going to proliferate, they will need tools that compress deployment timelines and reduce operational complexity. Netris is positioning itself as one of those tools.
How the money will be used
Netris plans to use the Series A funding to grow its engineering and sales teams, expand support for additional hardware vendors and add more capability to its automation algorithms. Those priorities point to both product depth and commercial scale.
More engineers should help the company keep pace with complex networking environments and continue adapting its software to new switch and server configurations. More sales capacity will be needed if Netris wants to move beyond a handful of early adopters and into wider deployment across the fast-growing neocloud segment.
Expanding vendor support is especially important because data center operators often mix equipment from multiple suppliers. A software platform that works across brands and architectures has a stronger chance of becoming a standard layer rather than a niche tool.
Inside the neocloud problem Netris is trying to solve
Neoclouds are attractive because they promise direct access to AI compute without the complexity of building entire cloud platforms from scratch. But the economics are unforgiving. Operators have to purchase costly GPUs, connect them through high-performance switches, and keep the entire stack humming while serving customers with different security and throughput needs.
That creates a set of operational challenges that can quickly become expensive:
- Hardware arrives before the environment is fully configured.
- Network teams must map, segment and validate thousands of connections.
- Customers want isolated compute environments for their workloads.
- Any delay increases the cost of idle GPUs and postponed revenue.
Netris is essentially targeting the middle of that chain. If it can help operators get a cluster live faster, manage it more efficiently and adjust it without large manual interventions, then the software becomes part of the economic engine of the neocloud business model.
How it compares with traditional cloud operations
Large hyperscale operators have spent years building internal tooling to solve these problems. They also have the scale to justify enormous engineering investments. Smaller AI infrastructure startups do not.
That is why automation platforms aimed at GPU networking may have a particularly large role to play in this market. In effect, they let a smaller company borrow some of the operational sophistication of a much larger cloud provider without having to build everything from scratch.
The difference is not just about convenience. It is about whether a new AI infrastructure company can keep its margins intact while moving fast enough to compete.
Key facts about the deal and the company
| Item | Details |
|---|---|
| Company | Netris |
| Funding round | Series A |
| Amount raised | $15 million |
| Lead investor | Andreessen Horowitz |
| Board addition | Guido Appenzeller, a16z partner |
| Primary use case | Automating setup and operations for GPU clusters and neoclouds |
| Reported deployment | More than 35 GPU clusters worldwide |
| Reported scale | About 1 million GPUs |
| Notable users | Lightning AI, Foxconn, Visionbay, HPE, Tensorwave, Telus |
| Hardware compatibility | Vendor-agnostic, including Nvidia and AMD ecosystems |
Why this matters beyond one startup
Netris’s funding is a reminder that the AI boom is no longer only about models, chips and cloud credits. The ecosystem is now mature enough that companies are building businesses around the operational friction created by AI infrastructure itself.
That matters because every major technology shift eventually produces a stack of supporting tools. In the early internet era, the winners were not just the websites; they were also the firms that enabled hosting, routing, load balancing and content delivery. A similar pattern is emerging in AI, where the bottlenecks have moved deeper into networking, orchestration and cluster management.
If neoclouds are to become a durable part of the AI economy, they will need software that makes their facilities easier to launch, safer to operate and cheaper to scale. Netris believes it has built one of those layers.
The next test will be whether the startup can turn technical credibility into broader market adoption while the AI infrastructure boom is still expanding. With fresh capital, a prominent investor and a product aimed at a real pain point, Netris now has a stronger shot at becoming one of the behind-the-scenes companies shaping the AI cloud race.
Timeline of Netris’s rise
| Period | Milestone | Significance |
|---|---|---|
| Years before the AI boom | Netris develops automation algorithms for network operations | Product foundation is built before current AI demand surges |
| About two years ago | Nvidia sees a demo and recommends the startup to customers | Early ecosystem validation and market exposure |
| Current stage | Deployment across 35+ GPU clusters globally | Evidence of traction at meaningful scale |
| Now | Raises $15 million Series A led by a16z | Capital to expand product, hiring and sales |
The bottom line
As AI infrastructure becomes more crowded and expensive, the companies that can reduce deployment time and operational complexity are gaining strategic importance. Netris is betting that the future of neoclouds will depend not only on securing GPUs, but on automating the network systems that make those GPUs usable at scale.
That is a narrower story than the one being told by many AI startups, but it may be a more practical one. In a market where every week of delay can leave valuable chips sitting idle, speed to live is becoming a competitive advantage in its own right.
