Anthropic is looking to grow its network engineer team with the Claude developer admitting its footprint is “growing faster than our ability to reason about it.”

Job listings for network engineers see the AI lab reveal it’s “turning up tens-of-terabits of private backbone capacity, peering across clouds, and moving model weights that keep getting larger.”

Anthropic is looking for engineers to bolster its Capacity & Efficiency team, which holds responsibility for its network, compute, and storage backbone. That team works alongside its Systems Networking team, which builds and operates its fabric, along with its Observability team that own the telemetry platform. The former are tasked with determining how bandwidth, latency, and dollars are being used, and unearthing potential optimizations.

The latest job description reveals Anthropic runs a private multicloud backbone built from dark fiber, optical transport, cloud solution provider (CSP) direct-connect products, “layered over data center fabrics spanning tens-of-thousands of hosts.” The description highlights the network's sheer size and costs, stating: “The scale is real, the spend is large, and the efficiency levers are still mostly unpulled.”

Anthropic wants engineers with backgrounds in cloud provider networking teams or cloud networking product teams, as well as familiarity with AI infrastructure traffic patterns and high-performance computing (HPC) fabrics.

Anthropic is also after candidates with knowledge of fabrics like InfiniBand and lossless Ethernet, as well as custom high-radix topologies. Prospective engineers may also be expected to have prior experience with multicloud connectivity, including cross-cloud peering, private interconnect products, and associated billing models.

The job offer includes a starting salary of $320,000, and expects all staff to be in one of its offices “at least 25% of the time.”

Anthropic shows no signs of slowing

The folks behind Claude started life as a splinter group of former OpenAI staff unhappy with disagreements over AI safety and its then anchor deal with Microsoft.

Having gone on to secure multibillion-dollar investments from Amazon and Google, Anthropic’s founders now finds themselves challenging their former employer, culminating in CEO Dario Amodei and OpenAI CEO Sam Altman’s awkward hand-holding snub of one another during a photo session with Indian Prime Minister Narendra Modi.

In recent weeks, Anthropic has repeatedly hit the headlines as it simply hasn’t stopped revealing new offerings, including an agentic AI building kit; a powerful, yet potentially problematic model dubbed Mythos; and version 4.7 of its flagship Claude Opus model.

To fuel all of its growing product roster, Anthropic has rushed to secure hardware from across the industry, most recently Google’s custom tensor processing unit (TPU) silicon line through a deal with Broadcom. That adds to the late 2025 reveal that Anthropic was the mystery Broadcom customer behind a $10 billion custom-chip deal.

Multi-year compute deals have also been signed with CoreWeave, along with a joint Microsoft and Nvidia partnership, while it’s also the sole user of Amazon's Rainer compute cluster housing nearly 500,000 custom Trainium2 chips.

Reports surfaced earlier this week that Anthropic wants to join the custom silicon club with apparent plans for its own hardware line to power Claude.