Skip to content

Aria Networks

5 viewers · 30d

Total raised

$73.0M

+$73.0M this year

1 filing since 2026 · latest Equity filed

Cumulative raised
LAST ROUND
Equity · $73.0M
ROUNDS
1
INVESTORS
33
FOUNDED
2024
HQ
Palo Alto, CA
SECTOR
Infrastructure
EMPLOYEES
N/A
30D VIEWERS
5

AI overview

Updated

Aria Networks builds Ethernet networking hardware and software for AI data centers, aimed at the back-end networks that connect GPU clusters. Its switches use Broadcom merchant silicon with a hardened version of the open-source SONiC network operating system, layered with telemetry collection and AI-driven optimization. The company frames its approach as 'Deep Networking' and optimizes for token efficiency and model FLOP utilization rather than raw switch throughput. It was founded in 2025 in Palo Alto by Mansour Karam (previously founder of Apstra, acquired by Juniper) and CTO Subhachandra Chandra, formerly of Arista.

What sets it apart

A path-centric rather than switch-centric architecture: Aria extracts microsecond-resolution telemetry already present in modern switching ASICs and applies probabilistic AI methods to optimize end-to-end paths across an AI cluster, instead of relying on rule-based per-switch configuration. The stack is chip-agnostic, so operators can change AI accelerators without rebuilding the network.

Funding history

1 round
EquityApr 6, 2026
Form D
+$73.0M$73.0M total

Latest SEC filings

via EDGAR · CIK 0002093784
Form D · Apr 6, 2026View on EDGAR

Products

1 tracked

The Network that Thinks

AI data center networking platform

An AI-native networking system combining Aria's Ethernet switches (Broadcom Tomahawk 5 class silicon), a hardened SONiC network operating system, end-to-end telemetry, and intelligent agents that continuously tune the network. Announced generally available in April 2026.

  • Hardened SONiC network operating system on merchant silicon
  • Microsecond-resolution telemetry extracted from the switching ASIC
  • End-to-end path optimization across the cluster, not per-switch rules
  • AI agents applying probabilistic optimization in place of static rules

Links

WebsiteLinkedInSEC EDGAR filings