InterviewAlly

Funding & Valuations

Nexthop AI Raises $500M to Build Next-Gen Data Center Networks

AI networking startup Nexthop AI has raised $500 million in Series B funding at a $4.2 billion valuation, launching three new switches that process up to 102.4 Tbps for GPU-to-GPU communication in AI data centers.

March 12, 2026 · 4 min read · Source: SiliconANGLE

Nexthop AI · Data Center · Networking · Series B · Lightspeed · AI Infrastructure

High-tech data center with illuminated networking switches and fiber optic cables in blue lighting

Nexthop AI Secures $500M to Rewire AI Data Centers

Nexthop AI has closed a $500 million Series B round at a $4.2 billion valuation, led by Lightspeed Venture Partners with participation from Andreessen Horowitz and Altimeter Capital. The oversubscribed round makes Nexthop one of the most valuable pure-play AI infrastructure startups, positioning it squarely in the growing market for data center networking gear optimized for GPU clusters.

Founded in 2024 by Anshul Sadana, former chief operating officer at Arista Networks, Nexthop is building networking hardware from the ground up for AI workloads -- environments where thousands of GPUs need to exchange data at speeds that traditional enterprise switches were never designed to handle.

Three New Switches Built for GPU-to-GPU Communication

Alongside the funding, Nexthop launched three new AI-optimized networking switches. The NH-4010 processes 51.2 terabits per second and is claimed to be up to 20% more power-efficient than competing products. The NH-4220 doubles that capacity, while the flagship NH-5010 is designed for a new "Disaggregated Spine" architecture purpose-built for hyperscale AI clusters.

All three switches use Broadcom networking chips and support the RoCEv2 protocol, which enables direct GPU-to-GPU data transfer by bypassing CPUs entirely. They also include DCQCN congestion detection technology and can save several dozen megawatts per device through intelligent power management -- a meaningful reduction when multiplied across thousands of switches in a hyperscale facility.

Why AI Networking Is Suddenly a $650B Problem

The funding reflects a broader recognition that networking has become a critical bottleneck in AI infrastructure. As Andreessen Horowitz partner noted, "AI clusters are pushing data center networks to their limits, and networking is now central to overall system performance." Tech giants including Alphabet, Amazon, Meta, and Microsoft are expected to spend approximately $650 billion in 2026 on AI data centers and related infrastructure.

Nexthop's operating system, Nexthop NOS, is based on Microsoft's open-source SONiC platform with added cybersecurity optimizations and patch management services. The company also announced linear pluggable optics (LPOs and LROs) that reduce optical networking costs by eliminating the need for digital signal processors.

What This Means for Engineers and Job Seekers

The explosion in AI data center spending is creating acute demand for network engineers, hardware designers, and systems architects who understand GPU cluster communication. Nexthop's rapid growth from founding to $4.2 billion valuation in under two years illustrates how fast the AI infrastructure talent market is expanding, with specialized networking expertise commanding premium compensation across both startups and hyperscalers.