A semiconductor startup founded by former Intel engineers has raised $100 million to develop networking chips and software aimed at solving one of the biggest infrastructure problems in artificial intelligence: moving data efficiently between processors.
The company, Delos Data, is targeting AI data centers that increasingly combine GPUs, CPUs, custom accelerators, memory and storage from different vendors. Its approach could create an alternative to tightly integrated networking systems built around NVIDIA’s hardware ecosystem.
However, the funding does not mean Delos Data has already challenged or displaced NVIDIA. The startup is still developing its technology and must prove that its networking products can perform reliably in large-scale commercial deployments.
Delos Data Raises $100 Million for AI Networking
Delos Data announced a $100 million funding round on September 15, 2026, according to Reuters.
The round included investments from:
- Matrix Partners
- Playground
- Socratic Partners
- Capricorn
- Matter Venture Partners
- IAG
- DYNAMIQ
Former Intel CEO Pat Gelsinger, who is now a partner at Playground, is also among the investors supporting the company.
Delos Data plans to use the funding to expand its hardware and software engineering teams, accelerate product development and support sales activities.
The company was founded by veterans of Intel with experience in data-center silicon and systems. Its leadership includes CEO and co-founder Ed Doe and CTO and co-founder Dan Daly.
Why AI Data Centers Need Better Networking
The AI infrastructure market has traditionally focused on building faster processors and adding more GPUs. NVIDIA has been a major beneficiary of this trend through its AI accelerators and networking technologies.
But faster processors are not useful if they spend too much time waiting for data.
Modern AI systems require constant communication between computing components. During model training and inference, processors exchange model parameters, activations, memory data and other information. Any delay in that communication can reduce utilization and increase energy costs.
This problem becomes more complex as data centers use a wider mix of hardware.
Instead of relying only on NVIDIA GPUs, operators may combine:
- NVIDIA GPUs
- AMD AI accelerators
- Cerebras systems
- CPUs
- Custom AI chips
- High-speed memory
- Storage and other data-processing components
Delos Data is building networking technology designed to help these different components communicate efficiently without being tied to a single type of computing hardware.
Delos Data’s Focus Is Data Movement
Delos Data’s strategy is based on the idea that the next major bottleneck in AI infrastructure may not always be processor performance. It may be the speed and efficiency of communication between processors.
The company is developing networking chips and software that aim to move data between different components with low latency and improved efficiency.
Its technology is intended to support data centers that use different kinds of accelerators and may change their hardware architecture over time.
Dan Daly, Delos Data’s CTO and co-founder, said the company does not know what the next infrastructure architecture for agentic AI will look like. He argued that Delos Data can help provide fast data movement regardless of how those systems evolve.
This is especially relevant as AI workloads shift from model training toward inference.
Training involves building or updating AI models, while inference involves running those models to answer questions, generate content or complete tasks. Agentic AI systems may perform longer sequences of actions, increasing the amount of communication required between computing resources.
The Startup Is Targeting NVIDIA’s Networking Ecosystem
NVIDIA’s influence in AI data centers extends beyond its GPUs. The company also provides networking products and software that help connect AI computing systems.
This integrated approach can make it easier for customers to build large AI clusters, but it can also create dependence on a particular hardware and software ecosystem.
Delos Data is targeting that area by developing networking solutions that can work across a more varied hardware environment.
The company’s pitch is that data-center operators should be able to combine different computing chips without allowing communication problems to reduce the value of their investment.
This could become increasingly important as cloud providers and large enterprises look for ways to control AI infrastructure costs, improve hardware utilization and avoid relying entirely on one supplier.
Still, Delos Data faces significant challenges. NVIDIA already has established products, software support, customer relationships and experience deploying large AI systems. A new networking-chip company must demonstrate not only strong technical performance but also compatibility, reliability, security and long-term support.
Delos Data Claims Major Efficiency Improvements
Delos Data has introduced its Nonstop AI Data Interface, alongside its previously announced Nonstop AI Cluster and Nonstop AI Server products.
The company says its interface can deliver:
- Up to 10 times lower latency
- Up to 10 times higher efficiency
These figures are company-reported performance claims. They should not be treated as independently verified industry benchmarks unless supported by detailed testing from customers, third-party laboratories or other independent evaluators.
Real-world performance will depend on factors such as:
- Network architecture
- Workload type
- Accelerator combination
- Software optimization
- Data-center scale
- Memory and storage configuration
- Deployment conditions
The key question is whether Delos Data can reproduce its claimed advantages in production AI environments rather than only in controlled demonstrations.
Why the Funding Matters for AI Infrastructure
The $100 million investment reflects growing interest in the infrastructure surrounding AI computing.
During the early stage of the generative AI boom, much of the industry’s attention focused on obtaining more powerful GPUs. As AI deployments become larger and more complex, data movement, memory access and networking are receiving more attention.
A networking startup does not need to replace every GPU to create an opportunity. It may instead focus on helping existing processors work more efficiently.
If successful, Delos Data could benefit from several long-term trends:
- More AI inference workloads
- Greater use of agentic AI systems
- Increased demand for heterogeneous computing
- Pressure to reduce data-center energy consumption
- Demand for better utilization of expensive accelerators
- Interest in reducing dependence on single-vendor infrastructure
The company’s technology could be relevant to cloud providers, AI laboratories, enterprise data centers and infrastructure operators that use multiple types of computing hardware.
Can Delos Data Actually Challenge NVIDIA?
Delos Data’s funding gives it capital to develop products and pursue customers, but it is too early to say that the startup has materially weakened NVIDIA’s data-center position.
NVIDIA’s advantage includes more than networking chips. Its broader ecosystem includes AI accelerators, interconnect technology, software libraries, development tools and relationships with major cloud and enterprise customers.
Delos Data is pursuing a narrower but important part of that infrastructure stack: communication between computing resources.
Its success will depend on whether it can deliver measurable benefits while fitting into existing data-center environments.
Important future indicators will include:
- Commercial product availability
- Confirmed customer deployments
- Independent performance testing
- Compatibility with major AI accelerators
- Production-scale reliability
- Software ecosystem support
- Additional funding or strategic partnerships
- Revenue generated from networking products
For now, the company should be viewed as a new competitor in AI networking infrastructure rather than an established replacement for NVIDIA.
The Bigger Shift in AI Data Centers
The Delos Data funding round highlights a broader change in how the AI hardware market is developing.
The industry is moving toward more diverse computing architectures. NVIDIA remains a major supplier, but AMD, Cerebras, cloud providers and custom-chip developers are also participating in the market.
As these systems become more diverse, networking technology may become more important.
The future of AI infrastructure may depend not only on which company builds the fastest accelerator, but also on which technologies allow different processors, memory systems and storage components to operate together efficiently.
Delos Data is betting that communication will be one of the most important parts of that future.
Conclusion
Delos Data, a startup founded by Intel veterans, has raised $100 million to develop AI networking chips and software for increasingly complex data centers.
Its goal is to improve communication between GPUs, CPUs and other accelerators while giving operators more flexibility in choosing their hardware.
The funding represents a significant step for the company, but it does not yet prove that Delos Data can displace NVIDIA or compete with its full AI infrastructure ecosystem.
The next stage will depend on product development, independent benchmarks and real-world customer deployments. If the startup can demonstrate reliable improvements in latency, efficiency and hardware interoperability, it could become an important participant in the rapidly evolving AI networking market.
Read More:- DeepSeek, Kimi and Qwen Narrow the AI Gap With the US
FAQ:-
What is Delos Data?
Delos Data is an AI infrastructure startup founded by Intel veterans. It develops networking chips and software designed to improve data movement inside AI data centers.
How much funding did Delos Data raise?
Delos Data announced a $100 million funding round in September 2026.
What problem is Delos Data trying to solve?
The company is trying to reduce communication delays and inefficiencies between GPUs, CPUs, accelerators, memory and storage in AI data centers.
Is Delos Data replacing NVIDIA?
No. Delos Data is developing networking technology that could provide an alternative or complementary solution to existing AI infrastructure. There is no verified evidence that it has replaced NVIDIA’s networking ecosystem.
What is NVIDIA’s role in AI networking?
NVIDIA provides AI accelerators and networking technologies used to connect and operate large AI computing systems. Its networking ecosystem is part of its broader data-center platform.
What products has Delos Data introduced?
Delos Data has introduced the Nonstop AI Data Interface, Nonstop AI Cluster and Nonstop AI Server products.
Are Delos Data’s performance claims independently verified?
The reported claims of up to 10 times lower latency and 10 times higher efficiency are company claims. Independent production-scale validation has not been established in the available reporting.
Why is AI networking becoming more important?
AI data centers increasingly use different types of processors and accelerators. Efficient communication between those components can help reduce idle hardware, improve utilization and control energy costs.
