Delos Data raises $100 million to develop chips for AI networks

Delos Data, a semiconductor startup founded by former Intel executives and engineers, has raised $100 million in new funding to develop chips and software aimed at improving data movement inside increasingly complex artificial intelligence data centers, Reuters reported Tuesday.

The investment comes as the AI hardware industry expands beyond traditional graphics processors and focuses increasingly on the technologies required to connect large numbers of chips efficiently.

Delos Data is developing networking solutions designed to allow different types of computing hardware to communicate more effectively. The company believes its technology could help data centers manage the rapidly increasing volume of information exchanged between processors as AI models become larger and more computationally demanding.

The startup was co-founded by former Intel veterans, including Dan Daly, who serves as its chief technology officer. Daly said Delos Data is focused on making communication across AI infrastructure faster and more adaptable as companies build increasingly large computing systems.

Efficient communication between processors has become an important challenge for AI data centers. As thousands of chips work together on training and running advanced AI models, moving data between those processors can become a significant factor in overall system performance, energy use and operating costs.

Former Intel CEO Pat Gelsinger, who is also a partner at venture capital firm Playground, participated in the latest financing round. Gelsinger said inefficient communication between chips can add to both energy consumption and expenses, underscoring the importance of networking technology as AI infrastructure continues to scale.

The funding round included Matrix Partners, Playground, Socratic Partners, Capricorn, Matter Venture Partners, IAG and DYNAMIQ, according to Reuters.

Delos Data enters the market at a time when Nvidia continues to hold a dominant position in AI computing and networking. Nvidia's GPUs are widely used for AI workloads, while the company has also developed networking technologies and other components designed to connect the large computing clusters required by modern AI systems.

However, the expanding AI market has created opportunities for other semiconductor companies. AMD and Cerebras Systems are among the companies competing in AI computing, while a growing number of startups are targeting specialized parts of the infrastructure stack.

Rather than competing solely on processor performance, companies such as Delos Data are targeting the communication layer that allows processors to operate together. This part of AI infrastructure is becoming increasingly important as data centers combine larger numbers of accelerators and other computing components.

The need for faster communication is closely linked to the rapid growth of AI workloads. Training sophisticated models can require enormous amounts of data to move between processors, while AI inference services must handle large numbers of requests with minimal delays.

Networking bottlenecks can limit the benefits of adding more processors if the chips cannot exchange information quickly enough. Improving that communication could therefore help data center operators make better use of existing computing resources.

Energy efficiency is another factor driving interest in the technology. AI data centers consume significant amounts of electricity, and the infrastructure required to transfer data between processors can contribute to overall power consumption.

Delos Data's approach is intended to support different combinations of computing hardware rather than being restricted to one particular chip architecture. That flexibility could become increasingly relevant as data center operators use processors and accelerators from multiple suppliers.

The company's latest funding also reflects continued investor interest in the infrastructure supporting the AI boom. Venture capital firms and other investors have committed substantial amounts of money to companies developing processors, networking systems, memory technologies and software for AI applications.

At the same time, the scale of AI infrastructure investment has generated questions about how long the current spending cycle can continue. Companies have announced major plans for new data centers and computing capacity, while investors have begun examining whether the pace of spending can be sustained.

Despite those concerns, demand for AI computing infrastructure remains strong as businesses deploy AI models for applications ranging from enterprise software and search to scientific research and automation.

For Delos Data, the opportunity lies in addressing an infrastructure challenge that can be less visible than processor performance but is becoming increasingly important: the efficient movement of information between chips.

As AI data centers grow from large clusters into even more interconnected computing systems, the ability to transfer data quickly and efficiently could become a key factor in determining how effectively those facilities operate.

The company's $100 million financing gives it additional resources to develop its chip and software technology as competition expands across the AI infrastructure market. Its progress will also be closely watched as semiconductor companies and startups seek to capture a larger share of the rapidly developing AI data center ecosystem.