Skip to content
AIPressRoom

FundingFoundation Models & Infrastructure

Firmus Lands $300M to Secure Trans-Pacific AI Connectivity

Firmus landed $300M to secure dedicated trans-Pacific connectivity for AI training and inference.

Firmus Technologies was co-founded by Oliver Curtis, Tim Rosenfield, and Jonathan Levee

The big picture: Firmus, an Australian AI firm, has committed $300M to a deal securing dedicated trans-Pacific connectivity for its AI operations.

Why it matters:

  • AI Infrastructure: Securing dedicated trans-Pacific connectivity is critical for high-performance AI training and inference, reducing latency and increasing data throughput.
  • Global Reach: This investment positions Firmus to expand its AI capabilities globally, leveraging distributed data centers and cloud resources more effectively.
  • Competitive Edge: Proprietary, high-bandwidth connections can offer a significant advantage in a competitive AI landscape where data transfer speeds directly impact model development and deployment.

How it works:

  • Dedicated Bandwidth: The deal ensures Firmus has exclusive access to a portion of the trans-Pacific cable’s capacity, bypassing public internet congestion.
  • Low Latency: Optimized routing and direct connections minimize the time required for data to travel between continents, crucial for real-time AI applications.
  • Scalable Operations: This infrastructure supports the massive data volumes and computational demands of advanced AI models, enabling Firmus to scale its training and inference workloads.

The catch: The substantial investment in physical infrastructure could tie up capital that might otherwise be used for AI R&D or talent acquisition. While ensuring connectivity, the long-term value depends on the sustained demand for such dedicated lines and the evolving landscape of global network infrastructure.

Key Facts

In this story

AIPressRoom Weekly

Every round we reported this week, the people behind the companies, and the events coming up. One email, every Tuesday.

More in Foundation Models & Infrastructure

All