Corning will manufacture Microsoft’s Hollow Core Fiber for Azure at scale, adding a major industrial partner to the cloud giant’s bid to remake the physical underpinnings of low‑latency AI and cloud networking worldwide.
Microsoft’s effort to commercialize hollow core fiber (HCF) for Azure began with its acquisition of Lumenisity in December 2022 and has accelerated through internal research and field deployments that the company says already carry live traffic. Microsoft’s engineering teams published progress reports and deployment notes that describe metro interconnect installations, performance targets and the ecosystem work required to make HCF operate in production networks.
In September 2025, Corning announced a strategic manufacturing collaboration with Microsoft to accelerate production of Microsoft’s HCF, confirming that Corning’s fiber and cable manufacturing facilities in North Carolina will produce the new fiber and related cable and connectivity products for Azure’s global network. The partnership focuses on operationalizing Microsoft’s manufacturing processes and creating next‑generation manufacturing platforms to scale yields, performance and reliability for mass deployment.
These developments come against a backdrop of a major technical breakthrough reported in peer‑reviewed literature and widely covered in the press: a Microsoft‑backed research team working with University of Southampton spinouts demonstrated a hollow‑core design with measured attenuation below 0.1 dB/km at 1,550 nm — a level that undercuts the long‑standing attenuation floor for conventional silica fiber and makes HCF viable for long spans and production networks. That work also reported multi‑dozen terahertz bandwidth and a roughly 45–47% improvement in effective photon transit speed compared with solid glass fibers.
For Azure customers running distributed AI training, model sharding, real‑time inference, and latency‑sensitive cloud services, reduced fiber propagation delay converts directly into faster sync times between datacenters and lower end‑to‑end latency for multi‑region requests.
Lower attenuation can reduce equipment count (fewer EDFAs, fewer repeaters), operational complexity, and power consumption — all material advantages for hyperscalers managing multi‑continental backbones.
The Nature Photonics‑reported experimental work that hit 0.091 dB/km was validated by tests on multi‑kilometre spools and multiple lab runs, showing low loss across wide spectral ranges and indicating the design is robust enough for further engineering toward production spools. Those results were published and discussed widely within the industry, catalyzing interest from system integrators and fiber manufacturers.
Corning’s announcement adds a manufacturing partner designed to take the next step — turning those validated designs into cable assemblies and connectivity products that can be installed at Azure scale.
For Azure customers and the broader Windows ecosystem, the most immediate advantages will appear on latency‑sensitive backbones and high‑value AI corridors where cloud operators can tightly control both ends of the optical path. Over the medium term, successful industrialization of HCF promises significantly faster inter‑datacenter links, higher capacity per fiber, and new operational models for cloud networks. The Corning–Microsoft partnership puts the industry one step closer to that future, but widespread adoption will be a careful engineering and economic journey rather than an overnight transformation.
Source: The Fast Mode Corning Expands Role in Microsoft Azure’s Global Fiber Infrastructure with Hollow Core Fiber
Background
Microsoft’s effort to commercialize hollow core fiber (HCF) for Azure began with its acquisition of Lumenisity in December 2022 and has accelerated through internal research and field deployments that the company says already carry live traffic. Microsoft’s engineering teams published progress reports and deployment notes that describe metro interconnect installations, performance targets and the ecosystem work required to make HCF operate in production networks. In September 2025, Corning announced a strategic manufacturing collaboration with Microsoft to accelerate production of Microsoft’s HCF, confirming that Corning’s fiber and cable manufacturing facilities in North Carolina will produce the new fiber and related cable and connectivity products for Azure’s global network. The partnership focuses on operationalizing Microsoft’s manufacturing processes and creating next‑generation manufacturing platforms to scale yields, performance and reliability for mass deployment.
These developments come against a backdrop of a major technical breakthrough reported in peer‑reviewed literature and widely covered in the press: a Microsoft‑backed research team working with University of Southampton spinouts demonstrated a hollow‑core design with measured attenuation below 0.1 dB/km at 1,550 nm — a level that undercuts the long‑standing attenuation floor for conventional silica fiber and makes HCF viable for long spans and production networks. That work also reported multi‑dozen terahertz bandwidth and a roughly 45–47% improvement in effective photon transit speed compared with solid glass fibers.
What Corning’s role actually is
Manufacturing at scale — why it matters
Hollow core fiber is not merely a laboratory curiosity: to affect global cloud infrastructure it must be produced in large volumes and integrated into cable systems and connectivity hardware that meet telco and data‑center standards. Corning’s agreement with Microsoft commits its North Carolina fiber and cable facilities to produce Microsoft’s HCF and complementary cable/connectivity products. That means Corning will:- Translate Microsoft’s HCF design and process controls into high‑volume manufacturing.
- Integrate HCF into cable constructs suitable for duct, inter‑datacenter, and intracity deployment.
- Develop connectors, splices, and assembly methods compatible with Azure operational workflows.
Operationalization and continuous improvement
Corning’s announcement stresses collaboration with Azure engineers to operationalize Microsoft’s manufacturing processes and to design next‑generation manufacturing platforms. Practically, this will involve:- Process transfer: converting lab recipes into reproducible manufacturing recipes with yield targets.
- Tooling and automation engineering: building lines that can deliver consistent geometry and thin‑membrane tolerances for HCF.
- Test and quality systems: creating inline optical metrology to detect microbends, surface defects, gas inclusions and other loss mechanisms unique to HCF.
- Connectorization and cabling: integrating HCF into real‑world cable armor, ribbon and loose‑tube constructs and compatible connectors.
The technical promise — what HCF changes and why Azure is betting on it
Latency and photon transit speed
HCF guides light primarily through air, not glass. Because the refractive index of air is close to 1.0 while silica is roughly 1.44–1.5, light in HCF travels closer to the speed of light in vacuum, reducing propagation delay per kilometer. Microsoft and collaborators report an effective photon transit advantage in the range of around 45–47% faster than standard silica single‑mode fiber under comparable conditions. That percentage represents the physical limit improvement available by replacing glass with air in the core and is the headline benefit for latency‑sensitive applications.For Azure customers running distributed AI training, model sharding, real‑time inference, and latency‑sensitive cloud services, reduced fiber propagation delay converts directly into faster sync times between datacenters and lower end‑to‑end latency for multi‑region requests.
Attenuation and reach
A major barrier for HCF historically has been elevated attenuation relative to silica. The recent DNANF (double nested antiresonant nodeless fiber) results reported attenuation below 0.1 dB/km at 1,550 nm, a practical threshold that makes long unamplified spans feasible and competitive with the best silica fibers. That breakthrough makes HCF plausible for metro and long‑haul segments without the same density of amplifiers or repeaters required before.Lower attenuation can reduce equipment count (fewer EDFAs, fewer repeaters), operational complexity, and power consumption — all material advantages for hyperscalers managing multi‑continental backbones.
Bandwidth, nonlinearities and high‑power launch
HCF designs report broader usable spectral bandwidth and dramatically reduced nonlinearities because most of the optical energy travels in air, not glass. The practical consequences include:- Wider usable wavelength windows, enabling denser wavelength division multiplexing (WDM) and new amplifier strategies.
- Higher launch powers without triggering nonlinear impairments such as self‑phase modulation and four‑wave mixing, yielding potential capacity increases per fiber pair.
- Improved signal integrity for coherent modulations used in modern high‑capacity links.
The deployment story so far — what’s been achieved and what’s planned
Microsoft’s engineers published details on trials and a metro interconnect deployment that they described as larger than previous HCF installations, and the company has reported pilot deployments carrying live traffic over more than 1,000 kilometers. Microsoft has also publicly announced ambitious rollout targets in past presentations and keynotes: public statements have referenced plans to scale to many thousands of kilometers within Azure’s backbone.The Nature Photonics‑reported experimental work that hit 0.091 dB/km was validated by tests on multi‑kilometre spools and multiple lab runs, showing low loss across wide spectral ranges and indicating the design is robust enough for further engineering toward production spools. Those results were published and discussed widely within the industry, catalyzing interest from system integrators and fiber manufacturers.
Corning’s announcement adds a manufacturing partner designed to take the next step — turning those validated designs into cable assemblies and connectivity products that can be installed at Azure scale.
Strengths of the Corning–Microsoft pairing
- Scale and industrial know‑how: Corning’s existing optical cable campus and automation experience dramatically lower the barrier to scaling HCF. Turning a lab fiber into millions of kilometers of field‑rated cable requires factory expertise that Corning has cultivated for decades.
- Vertical integration for the cloud: Microsoft’s control over the fiber design (through Lumenisity and its internal research) combined with Corning’s manufacturing and cable expertise is the classic producer‑partner model for cloud infrastructure — co‑designing components to ensure end‑to‑end performance and reliability.
- Faster path to market: By leveraging an established manufacturer in a country with skilled fiber technicians and supply‑chain links, Azure can accelerate the timeline from prototype to field deployment.
- End‑to‑end productization: Corning’s involvement extends beyond raw fiber drawing to cable and connectivity products, which simplifies the supply chain for Azure and reduces integration risk during installation and maintenance.
Risks, unanswered questions and engineering hurdles
While the technical promise is real, several pragmatic risks and operational questions remain before HCF becomes ubiquitous.1. Manufacturing yield and geometric tolerance
HCF’s guidance mechanism depends on microstructured glass membranes and precise geometries; small deviations can significantly change loss and modal behavior. Scaling yields to cloud‑grade volumes requires tight process control and high first‑pass yields. Corning’s manufacturing expertise mitigates this risk, but the challenge remains substantial.2. Connectorization, splicing and field repair
HCF has unique mode profiles and alignment sensitivities. Existing splicing equipment and connectors that work for single‑mode silica fiber may not directly translate. Successful production requires validated field splicing processes, fusion splicer firmware or tooling updates, and training for field crews. These operational costs must be accounted for in large deployments.3. Bend sensitivity and conduit handling
Some HCF designs are more sensitive to tight bends and microbending. Cable designs and installation practices must be adjusted to protect the fiber during pulls, bends and repairs. Corning’s cable engineering will need to preserve fiber geometry while meeting conventional duct and tray specs.4. Ecosystem compatibility and standards
The WDM, amplifier and transceiver ecosystems are deeply tied to silica fiber behaviour. While HCF relaxes some constraints (e.g., higher launch power), it also introduces new variables for amplifier design, Raman considerations, and optical power budgets. Interoperability testing across vendors and the development of industry standards will be crucial for multi‑vendor networks and third‑party carriers. The effort to standardize HCF in telecom specs is ongoing.5. Economic tradeoffs and total cost of ownership
HCF may command a price premium per kilometer during early production. The total cost of ownership calculus must include fiber cost, cable packaging, specialized connectors, training, and potential savings from fewer amplifiers and lower latency. Cloud providers are in a unique position to amortize early costs because they operate at hyperscale, but the timeline for cost parity with silica remains uncertain.6. Long‑term reliability and environmental resilience
Networks run for decades; field‑proven long‑term reliability over varied temperatures, humidity and mechanical stress must be demonstrated. Lab results are promising, but real‑world deployments on installed ducts and undersea systems will reveal long‑term failure modes.What this means for Azure customers and Windows ecosystem users
For enterprise cloud customers and AI workloads
- Lower inter‑datacenter latency: Faster photon transit and reduced propagation delay mean distributed model training and large dataset synchronization will be quicker. For multi‑region AI pipelines, cumulative latency reductions across many links can materially improve training and inference cycles.
- Higher capacity and fewer amplifiers: Potentially higher spectral efficiency and reach can reduce the need for repeater equipment, simplify network topologies and decrease operational power consumption.
- New classes of applications: Lower latency and higher throughput may broaden the feasibility of remote robotics, surgical telepresence, high‑frequency trading and real‑time collaborative simulations hosted in the cloud.
For Windows users, gamers and VDI customers
- Improved cloud gaming and streaming: Reduced backbone latency can shave tens of milliseconds from long‑distance connections, which compounds with last‑mile improvements to enhance responsiveness for cloud gaming and remote desktop services.
- Smoother remote work experiences: Virtual Desktop Infrastructure (VDI) and Windows‑hosted remote apps that span regions could see improved interactivity when application state and storage synchronizations are faster.
Critical analysis — realistic timelines and expectations
The research community’s 0.091 dB/km result and the Microsoft trials are watershed moments for HCF, but laboratory metrics rarely translate to instant global rollouts. Expect a staged, multi‑year transition:- Pilot and metro deployments to validate installation, connectorization and field repair processes.
- Incremental integration into high‑value routes where latency matters most (AI backbones, financial corridors, research networks).
- Gradual production scaling as manufacturers like Corning optimize yields and reduce per‑meter cost.
- Broader adoption once standards mature and the ecosystem of splicers, transceivers and cable hardware are fully validated.
Competitive and strategic implications
- For Corning: The partnership deepens a longtime relationship with a hyperscaler and positions Corning as a primary supplier for next‑generation fiber optics. Success here could open new product lines and cement Corning’s role in cloud backbone engineering.
- For other fiber vendors and telcos: The Corning–Microsoft play signals an industry shift; other manufacturers and carriers will accelerate their HCF R&D and pilot programs to avoid becoming dependent on a single supply source for high‑performance fiber.
- For standards bodies and integrators: Rapid progress forces bodies and vendors to accelerate standardization, interoperability testing and the creation of best practices for installation and maintenance.
- Geopolitics and supply chain: Large‑scale HCF deployment implies geopolitical leverage over key backbone infrastructure. Nations and operators will weigh sourcing strategies, local manufacturing and vendor diversification to avoid strategic dependencies.
Practical checklist for network architects and IT leaders
- Audit which backbone routes and inter‑region links in your cloud topology are latency‑sensitive and could benefit from HCF.
- Engage cloud vendors to understand HCF availability timelines for the regions you use.
- Factor potential HCF adoption into long‑term optical planning; anticipate new procurement items (connectors, splicing services, test equipment).
- Plan for transitional operational training for network field crews should your organization ever own or manage lines that include HCF segments.
- Maintain expectations: initial benefits will be in the backbone; last‑mile and access networks will continue to dominate perceived user latency for many applications.
Looking ahead — what to watch for
- Manufacturer yield updates and per‑meter cost reductions from Corning’s North Carolina lines.
- Field interoperability tests between HCF segments and conventional silica fiber systems, particularly across vendor boundaries.
- Standards activity defining connectors, splicing methods, and optical power budgets for HCF in telco standards forums.
- Performance reports from Azure about latency and throughput improvements on production HCF routes.
- Broader industry adoption stories from carriers and other hyperscalers — these will indicate whether HCF becomes a multi‑vendor industry trend or remains a hyperscaler differentiator.
Conclusion
Corning’s move to manufacture Microsoft’s Hollow Core Fiber for Azure is a pivotal step in turning a long‑promised optical technology into an operational reality. The combination of Microsoft’s design and field experience, Corning’s manufacturing scale, and the recent record‑breaking optical results collectively reduce the technical risk of HCF adoption. At the same time, practical hurdles — yields, splicing, connectorization, standards, and economics — remain and will determine the pace and breadth of deployment.For Azure customers and the broader Windows ecosystem, the most immediate advantages will appear on latency‑sensitive backbones and high‑value AI corridors where cloud operators can tightly control both ends of the optical path. Over the medium term, successful industrialization of HCF promises significantly faster inter‑datacenter links, higher capacity per fiber, and new operational models for cloud networks. The Corning–Microsoft partnership puts the industry one step closer to that future, but widespread adoption will be a careful engineering and economic journey rather than an overnight transformation.
Source: The Fast Mode Corning Expands Role in Microsoft Azure’s Global Fiber Infrastructure with Hollow Core Fiber
