Delos Data Inc., a startup that develops hardware for artificial intelligence clusters, has raised more than $100 million in funding.
The capital was provided by Matrix, Playground, Socratic Partners, Capricorn, Matter Venture Partners and IAG. Delos stated in its Tuesday funding announcement that a group of angel investors from the data center hardware industry chipped in as well.
The raise comes three months after the company debuted its first product. The Delos Nonstop AI Server is a computing appliance optimized for inference. Customers can equip it with up to four graphics cards from external suppliers. The accelerators attach to the system’s motherboard via an open-source interconnect called OAM that is specifically optimized for AI use cases.
Customers can link together multiple Nonstop AI Servers into a cluster. The machines exchange data via OSFP, an open-source routing technology. Each of Nonstop AI Server’s four chips has nine OSFP ports with 1.6 terabits per second of bandwidth apiece. That reportedly gives the system a comparable amount of network capacity to Nvidia Corp.’s Rubin appliances.
Delos debuted a new addition to its hardware lineup in conjunction with its funding announcement. The Delos Nonstop AI Data Interface, as it’s called, is a chip designed to work with the Delos Nonstop AI Server. It enables the server to connect to the network of the inference cluster in which it’s deployed.
The company says that its new networking chip provides 10 times lower latency than rival products. Furthermore, Delos claims that it’s more efficient and scalable.
In addition to powering network connections, the chip monitors the host cluster for malfunctions. It detects issues such as graphics card crashes and network link outages. The Nonstop AI Data Interface also automates some of the work involved in recovering from such failures.
Delos sells the technology in three flavors. The first is a PCIe card that data center operators can attach to their servers. The other two implementations, in turn, enable customers to integrate the technology into their systems at the component level.
It offers a Nonstop AI Data Interface chiplet that can process more than 30 terabits of traffic per second. Customers can integrate it directly into their AI accelerators. There’s also a near-packaged optics, or NPO, version with a capacity of over 10 terabits per second. It’s designed to be placed on the same substrate as the accelerator that uses it. That arrangement provides lower speeds than a chiplet integrated directly into the accelerator but can be simpler to implement.
“Agentic AI is scaling beyond a single device, and when that happens performance isn’t solely determined by the processors, it’s decided by how the data moves between them,” said Delos co-founder and Chief Technology Officer Dan Daly (pictured, right, with co-founder and Chief Executive Officer Ed Doe). “Agentic AI needs a data interface that’s fast and doesn’t break under load.”
Delos will use its newly closed funding round to hire more engineers and accelerate its sales efforts.
