Key takeaways

  • Flag announced a corporate rebrand in April 2025, reverting to the historic “Flag” name.
  • $250 million of post‑acquisition capital is slated for network upgrades and AI‑ready services through the end of 2026.
  • The subsea system stretches 66,000 km, currently delivering about 90 % of revenue from subsea and 10 % from terrestrial services; the target is to shift toward a 60 % / 40 % split.
  • AI workloads are reported to demand roughly ten times the bandwidth of traditional cloud traffic, accelerating demand for low‑latency paths.
  • Flag bought a fiber pair on Google’s Echo cable, creating a Singapore ↔ U.S. Pacific corridor that underpins AI‑intensive data‑center traffic.
  • New terrestrial fiber projects are under way in Iraq and Saudi Arabia to add redundancy for regional data‑center clusters.

Rebranding to Flag

In April 2025 the company announced a name change, abandoning the Global Cloud Xchange (GCX) brand and returning to the flagship “Flag” label that industry veterans still recognize. The move was presented as part of a broader revival that includes modernising the digital platform and sharpening the focus on AI‑driven customers.

Private‑equity backing and capital deployment

Flag’s turnaround was accelerated by a $512 million purchase by 3i Infrastructure in September 2022. The private‑equity sponsor has earmarked a quarter‑billion dollars for network expansion, with the spend expected to be fully deployed by the close of 2026. The funding is directed at optical‑layer upgrades, AI‑focused telemetry and the addition of both subsea and terrestrial capacity.

Scale of the subsea footprint and service mix shift

Flag’s undersea infrastructure spans more than 66,000 km (41,010 mi) and links North America with Asia across a portfolio of systems that include FA‑1, HAWK, Falcon, MEDEX, FEA, IEX and FNAL. At present roughly nine‑tenths of the company’s revenue is derived from subsea services while the remaining one‑tenth comes from terrestrial offerings. Management is deliberately tilting that balance toward a roughly 60 % subsea and 40 % terrestrial split to capture growing demand from data‑center clusters in the Middle East.

Pacific corridor for AI traffic

A concrete illustration of the Pacific emphasis is Flag’s recent acquisition of a fiber pair on Google’s Echo cable. The route runs from Singapore to the United States, landing at Eureka, California, with intermediate stops at Guam and Changi, Singapore. By securing capacity on this low‑latency corridor, Flag positions itself to serve AI‑intensive workloads that span the Pacific region.

Terrestrial expansion and route diversity

Flag is augmenting its terrestrial network with new fiber links in Iraq and Saudi Arabia. These projects are designed to give regional data‑center operators alternative paths should subsea cables be disrupted, and they also broaden the overall mix of land‑based capacity that the company can market.

AI‑driven bandwidth demand

According to the CEO, AI workloads now require roughly ten times the bandwidth that cloud applications did a few years ago. The surge is prompting an “arms race” among carriers to deliver higher capacity, tighter latency budgets and greater redundancy—all of which Flag is addressing through its investment program and Pacific‑focused route strategy.

Outlook

Flag’s rebrand, capital injection and strategic emphasis on Pacific connectivity aim to create a platform that could help meet the anticipated bandwidth growth generated by AI. By shifting a larger share of its business to terrestrial assets while still leveraging a massive subsea network, the company aims to offer resilient, low‑latency paths for the next wave of data‑center demand.

Conclusion Flag’s identity change is more than a marketing refresh; it signals a concrete pivot toward AI‑centric data‑center connectivity. With $250 million earmarked for upgrades, a Pacific corridor that directly serves AI traffic, and a deliberate move to increase terrestrial capacity, the company is aligning its asset base with the bandwidth and latency characteristics that AI workloads demand.

Sources

This article was researched and fact-checked against the following sources: