Summary
- Chunghwa Telecom, Fujitsu and 1Finity signed a two-year strategic partnership for technical verification.
- The programme will use Chunghwa submarine-cable systems to test all-photonics networking.
- Distributed data centres are a feasibility and technical-verification target, not a deployed service.
- Quantum-technology exploration is framed around security needs for network control, not a commercial quantum product.
- The work continues cooperation begun in 2024; no purchase value, capex, capacity, customer, route or launch date was disclosed.
A submarine cable usually appears in a network plan as transport: a fixed path that carries traffic between landing points. Chunghwa and Fujitsu want to use that infrastructure as a test bench for something more programmable. Their question is whether end-to-end optical paths can be controlled across cable systems and distributed computing locations with less repeated electrical conversion.
That is an engineering proposition, not yet an operating product. The parties have committed two years to verification. The duration makes the programme more substantial than a demonstration-day announcement, while its missing commercial terms keep it on the research side of the investment boundary.
The cable supplies a real constraint
Laboratory optical links can prove component performance. An operator cable system adds route distance, existing equipment, protection arrangements, maintenance practice and live-network control. Tests on that surface can reveal whether an APN design integrates with the systems that actually carry international connectivity.
The announcement does not name a route or promise that production traffic will be moved onto a new architecture. It also gives no optical capacity, equipment quantity or performance target. The useful evidence will therefore be measured results: provisioning time, path stability, recovery behaviour, power use and interoperability with the current network.
1Finity participates as an optical-networking specialist, while Fujitsu and Chunghwa extend work that began in 2024. That history matters because the new agreement is a next verification phase rather than a claim that the earlier phase delivered nationwide implementation.
Distributed data centres remain a feasibility question
An all-photonics path can matter to distributed computing if it connects facilities with predictable latency and sufficient bandwidth. Workloads could be placed across several sites while the optical layer is configured around demand. The commercial value would come from useful computing delivered, not from the existence of a photonic link alone.
The agreement calls for feasibility verification of distributed data centres. It does not guarantee that facilities will be built, name their locations or disclose tenants and power. It also does not say conventional packet and electrical networks will be replaced. Those layers still handle functions that an optical path announcement does not remove.
The two-year programme should establish where optical continuity helps and where conversion, routing, orchestration or fault isolation remains necessary. A negative result can be useful if it prevents capital from being committed to a topology that is difficult to operate.
Quantum work begins with a security need
The partners will explore quantum technology in relation to security for network control. That wording matters. It describes research into how control systems may be protected as threats and cryptographic requirements change; it does not announce a quantum-safe commercial service.
No algorithm, hardware deployment, customer or certification is specified. The credible milestones are threat models, test interfaces and evidence that security measures work without making network recovery or provisioning impractical.
The next decision needs economics
The partnership omits contract value, capex, order volume and a launch date. Those are not minor disclosure gaps when judging deployment. A technical result becomes an investment case only after it can be tied to equipment cost, operating complexity, energy, service demand and the value of faster or more flexible paths.
Over the next two years, the strongest proof would be a reproducible test across a named topology, measured against the existing network, followed by a bounded production trial. Until that appears, Chunghwa’s cable systems are a serious verification environment—not evidence that Taiwan has launched a nationwide APN.