Google’s network architecture provides one example of this shift. Its global infrastructure spans more than 10 million kilometres of terrestrial and subsea fibre and connects multiple cloud regions and edge locations. The company has also described its Virgo architecture as a means of expanding AI compute across multiple data centres. These developments illustrate how fibre networks can support connectivity across different layers of an AI environment, from individual facilities to geographically separated compute resources.

Distributed Compute Increasing Importance of Optical Connectivity

The changing geography of AI is creating requirements that extend across intra data centre, inter data centre, regional and international connectivity. ITU T’s ION 2030 framework identifies optical networking as an important foundation for distributed AI training, inference and data exchange, while operators are expanding fibre infrastructure around new AI and cloud facilities. Microsoft, Meta, Bell and Ooredoo provide examples of infrastructure investments linking AI data centres with regional and international fibre systems.

This development also connects with network power efficiency, because scaling distributed compute increases the importance of efficient network infrastructure alongside capacity. The central change is that fibre networks are increasingly being planned not only around access demand, but also around the geographic distribution of computing resources.

Distributed AI computing is expanding the role of fibre from user connectivity toward high capacity links connecting computing resources across locations.

The technical progression is also changing the relationship between data centre switching and optical transport. The OIF’s 800ZR framework provides an interoperable coherent interface for high capacity connectivity, while IEEE development work is advancing the specifications associated with 1.6TbE and 200G per lane signalling. These developments show that 800G and 1.6T optics are not simply higher speed versions of existing modules. They require changes across electrical interfaces, optical modules, signal processing and network architecture.

1.6T Extending the Optical Capacity Roadmap

The progression from 800G toward 1.6T is also being shaped by the need to carry more traffic across metropolitan, regional and longer reach data centre connections. Commercial and technical activity already spans Europe, Asia Pacific, India, the Middle East and other markets, with operators testing or deploying high capacity coherent solutions across live networks. These developments indicate that 800G and 1.6T optics are increasingly being considered as part of the wider evolution of data centre connectivity rather than as isolated technology demonstrations.

Higher optical capacity is increasing traffic per interface while making reach, power, signal integrity and interoperability more important to data centre network design.

Private connectivity services from major cloud providers also illustrate how enterprise requirements are evolving. AWS supports dedicated connections at up to 400 Gbps through Direct Connect, while Microsoft Azure ExpressRoute Direct offers 100 Gbps and 400 Gbps ports. Google Cloud made 400 Gbps Cloud Interconnect generally available in 2026. These capabilities do not indicate that every enterprise requires such capacity. Instead, they show how cloud connectivity is increasingly supported through dedicated network infrastructure alongside conventional internet access.

Optical Networks Adapting to Distributed Cloud Requirements

Operator networks are adapting by expanding optical capacity, cloud adjacency and geographic reach. Arelion reported that 90% of its European and US long haul network supported 400GE in 2025, with more than 250 points of presence offering 400GE. In Africa, Liquid Intelligent Technologies operates more than 116,000 kilometres of fibre and provides cloud connectivity through its terrestrial and international network. In Asia Pacific, PLDT has upgraded optical transport capacity to support traffic between international cable infrastructure, its national backbone and global cloud providers. Together, these examples show how high speed fibre is becoming part of a broader connectivity chain linking businesses with distributed digital infrastructure.

The development also changes the requirements placed on network operators. Capacity remains important, but enterprise and cloud customers can also require predictable performance, private connectivity and access across metropolitan, national and international routes. This is creating a broader role for high speed fibre as operators adapt networks around increasingly distributed enterprise and cloud architectures.

Rising interconnection requirements are expanding the role of fibre networks in connecting enterprises, cloud platforms and distributed digital infrastructure.

Private Connectivity Expanding Across Enterprise Networks

The growth of cloud based workloads is changing how enterprises approach connectivity between offices, data centres and external platforms. High speed fibre can provide dedicated paths for organisations that require predictable bandwidth and private connections rather than relying solely on shared internet access. AWS Direct Connect, Microsoft Azure ExpressRoute and Google Cloud Interconnect all support dedicated connectivity between customer infrastructure and cloud environments, with capacity options extending into hundreds of gigabits per second. These services illustrate how cloud connectivity is becoming an integrated part of enterprise network planning.

The requirements also extend beyond bandwidth. Enterprises operating distributed applications can need consistent performance between multiple locations, while organisations with larger data transfers may require dedicated optical capacity. This is encouraging network operators to place greater emphasis on metro reach, long haul connectivity and direct access to cloud and interconnection facilities. High speed fibre is therefore becoming part of a wider enterprise architecture in which optical networks connect multiple points of digital activity rather than serving as a simple last mile link.

Optical Transport Networks Adapting to Cloud and Enterprise Workloads

Operators are responding by expanding fibre capacity and positioning networks closer to cloud and data centre ecosystems. Arelion reported that 90% of its European and US long haul network supported 400GE in 2025, with more than 250 points of presence offering 400GE. Liquid Intelligent Technologies operates more than 116,000 kilometres of fibre across Africa and provides direct cloud connectivity through its terrestrial and international network. PLDT has also upgraded its optical transport infrastructure to support traffic between international cable systems, its national backbone and global cloud providers. These developments show how operators in different regions are adapting networks around increasingly distributed sources and destinations of enterprise traffic.

The resulting network model connects enterprise sites, cloud platforms, data centres and international infrastructure through multiple fibre layers. This places greater emphasis on network reach, service reliability and the ability to provision capacity across different routes. High speed fibre also supports a broader range of wholesale services, including dedicated wavelengths and private cloud connections, allowing operators to serve customers with requirements that differ from conventional broadband.

As these networks become more closely integrated with enterprise IT environments, operators must also manage the relationship between physical fibre infrastructure and increasingly flexible service requirements. This is creating demand for connectivity that can be provisioned across metropolitan and long haul routes while maintaining defined performance characteristics. The same evolution is also opening the way toward higher capacity optical connectivity, where data centre interconnection requirements place additional demands on optical transport networks.

Under the cloud infrastructure agreement, Anthropic will use Akamai Cloud’s distributed infrastructure and software to support the expansion of its CPU workloads at scale. The arrangement strengthens Akamai’s position in providing infrastructure for large enterprise and AI workloads across its distributed cloud environment.

The agreement also includes provisions that could expand the relationship by up to an additional $9 billion during the seven-year term. This would bring the total potential commitment between the companies to approximately $20 billion, subject to further purchases of cloud services under mutually agreed terms.

As part of the expanded relationship, Akamai has issued Anthropic a warrant to purchase non-voting convertible Series B Preferred Stock representing 7.7 million Akamai shares on an as-converted basis. The warrant could represent up to approximately 5% of Akamai’s common stock outstanding.

Akamai Expands Distributed Cloud Infrastructure

The cloud infrastructure agreement is designed to support Anthropic’s increasing CPU workload requirements through Akamai Cloud’s distributed architecture. Akamai said its platform provides compute capabilities across core and edge environments through a network spanning thousands of points of presence.

The company has described its cloud platform as supporting a continuum of compute and a diversified hardware environment for building, deploying and operating applications. For AI workloads, Akamai said its infrastructure is designed to support application performance, reliability and security throughout the application lifecycle.

The arrangement with Anthropic adds to more than $2.8 billion in multi-year Cloud Infrastructure Services commitments across Akamai’s customer base that were announced earlier in 2026. The company said those agreements reflect increasing demand for infrastructure that can support customers building, deploying and operating AI workloads at scale.

The cloud infrastructure agreement therefore represents a major expansion of Akamai’s cloud services business, with the company preparing additional infrastructure capacity to accommodate Anthropic’s expected workload growth.

Akamai estimates that total capital expenditures associated with the $11.6 billion commitment will be approximately $5.5 billion. The company also expects capital expenditures to increase by approximately $1.7 billion in 2026 as it secures and pre-purchases critical supply chain components, including memory.

Anthropic Relationship Could Expand Further

The cloud infrastructure agreement includes a mechanism for the relationship to grow beyond the initial $11.6 billion commitment. Up to an additional $9 billion in cloud service purchases could be added within the seven-year term.

Under the arrangement, each additional $3 billion purchase of cloud services at mutually agreed terms would result in the vesting of approximately 1% of Akamai’s common stock outstanding under the warrant. Approximately 2% of Akamai’s common stock outstanding is expected to vest in connection with the initial commitment, while the remaining approximately 3% is linked to the potential expansion of the relationship.

Akamai said the agreement will support the company’s broader efforts to provide distributed infrastructure for AI applications and workloads. The company does not anticipate an impact on its 2026 revenue guidance from the commitment.

For Anthropic, the cloud infrastructure agreement provides access to Akamai Cloud’s distributed infrastructure and software to support CPU-intensive workloads. For Akamai, the arrangement expands its role in providing cloud capacity for large-scale AI applications.

The cloud infrastructure agreement establishes a long-term infrastructure relationship between the two companies, with the potential for further expansion as Anthropic’s computing requirements grow.

The upgraded cloud native 5G core is designed to deliver a planned capacity of 9 terabits per second, positioning it among Europe’s largest mobile packet cores. That level of throughput is critical for meeting the growing data demands driven by streaming, gaming, Fixed Wireless Access home broadband and other bandwidth-intensive services across VodafoneThree’s base of 28.6 million customers.

Deployed on Ericsson’s Cloud Native Infrastructure Solution, the new core provides a simplified, more resilient and automated network environment. By unifying 4G and 5G core functions on a single platform, the architecture reduces complexity while laying the groundwork for 5G Standalone capabilities and advanced services such as network slicing.

Differentiated Connectivity Through Network Slicing

The cloud native 5G core underpins VodafoneThree’s introduction of differentiated connectivity services, including the recently launched Vodafone SuperMobile. The service delivers up to four times faster speeds and a minimum download speed guarantee of 15 Mbps when customers are within 5G Standalone coverage.

Two dedicated network slices support these capabilities. The 5G+ FastTrack slice gives consumers access to a dedicated portion of the network, helping data move faster and more reliably during periods of high demand. The National Business Slice separates business traffic from consumer demand, enabling up to four times faster speeds and committed performance targets so that employees, applications and critical workflows remain connected and productive.

VodafoneThree’s network slicing strategy builds on the 5G+ Local Slicing service launched by Vodafone Business earlier in 2026. That offering delivers contracted performance at specific high-density sites such as entertainment venues, travel hubs and logistics facilities. Looking ahead, VodafoneThree plans to launch a National Critical Slice in 2027, designed to give blue-light organisations and critical public services priority connectivity and assured performance for mission-critical operations.

From Shared Infrastructure to a Dedicated Core

The upgraded cloud native 5G core represents the second major phase of VodafoneThree’s ambitious network transformation. Earlier in 2026, VodafoneThree completed a world-first network upgrade that unified core-level sharing through Multi-Operator Core Network technology with its existing Multi-Operator Radio Access Network across a multi-vendor 4G and 5G network. That step enabled Vodafone and Three customers to automatically connect to the best available coverage from either network at no extra cost. The current deployment advances that work from shared infrastructure to a single, dedicated and highly advanced Ericsson 5G Core.

The core network project was originally contracted by Three UK in January 2025 to support increasing data demand. It has since been seamlessly integrated into the wider VodafoneThree network strategy. Ericsson serves as VodafoneThree’s sole vendor for the nationwide core network and provides the significant majority of its radio access network, as part of the eight-year partnership announced in September 2025. The ongoing modernisation also includes deployment of Ericsson’s latest energy-efficient radio hardware and smart multi-band antennas across major population centres.

Hrvoje Jerkovic, Network Engineering Director at VodafoneThree, said: “The activation of our new cloud-native 5G core network is a major milestone in our journey to build the UK’s best network. Following the successful integration of our radio networks earlier this year, we are now taking the next step by creating a single, highly advanced core network that will underpin the future of connectivity for consumers and businesses across the UK. Together with Ericsson, we are building a network designed for the demands of tomorrow, delivering greater capacity, resilience and performance, while creating the foundation for innovations such as network slicing and new AI-powered services.”

The solution is intended to give operators faster access to trusted and unified data, reducing the time required to prepare information for operational use. Nokia said the integrated approach can make high-quality data available in minutes rather than weeks, addressing a key challenge for telecom providers seeking to scale telecom network automation across their network operations.

Unified Data Infrastructure Supports Network Automation

Nokia Data Suite provides prebuilt and reusable telco data products with data quality controls and telecom-specific semantic modelling. Microsoft Fabric contributes a unified data platform with OneLake storage, analytics tools and AI-enabled applications.

By combining the two platforms, telecom operators can connect network data with enterprise, IT and third-party information in a broader environment. The architecture is designed for multi-vendor and cross-domain networks and can support hybrid, cloud and on-premises infrastructure.

The companies said the approach can support use cases such as automated root-cause analysis and closed-loop operations. It also aims to reduce the data integration work traditionally required before AI applications can be introduced into network environments, supporting broader telecom network automation strategies.

The flexibility of the platform is intended to support operators that need to maintain regulatory requirements while expanding their use of AI and automation. Nokia and Microsoft are positioning the unified data foundation as an infrastructure layer for building agent-based solutions across different parts of the telecom network.

AI Use Cases Extend Across Telecom Network Operations

Initial use cases include autonomous Voice over New Radio, or VoNR, assurance, geo-experience and predictive maintenance and fault management. The VoNR capability is designed to identify service anomalies, conduct root-cause analysis and recommend actions using network, service and subscriber visibility.

The geo-experience capability combines subscriber, network and radio frequency data to map sessions to precise locations, identify degraded radio performance and locate coverage or capacity hotspots. AI agents can then analyse potential causes of radio access network issues and provide recommendations for corrective action across standard and network-sliced 4G and 5G services.

Predictive maintenance and fault management use historical and real-time data to identify potential network issues before they affect customers. These capabilities can contribute to telecom network automation by helping operational systems analyse conditions and support workflows with less manual intervention.

The collaboration also includes AI-driven operational assistants that provide network engineers with contextual insights, recommended actions and automated workflow execution. This approach is intended to support higher levels of telecom network automation while retaining human oversight within defined governance boundaries.

The solution is available now, with Nokia and Microsoft continuing to expand autonomous network use cases and customer deployments. The companies said the operating model is designed to allow AI systems to take on increasingly complex network responsibilities while maintaining human oversight.

The announcement, made on September 8, 2026, deepens a relationship between the two companies that stretches back three decades. Under the terms of the deal, Corning will supply the optical fiber deployment needed for Verizon to pursue its broadband and AI connectivity objectives over the next several years.

Broadband Expansion and Converged Architecture

A significant portion of the fiber covered under the Verizon and Corning fiber agreement is intended to accelerate Verizon’s broadband expansion across the United States. The company has set a target of reaching 40 million to 50 million broadband passings, extending fiber-to-the-home and local business connectivity to communities nationwide. It is important to note that this figure represents a company target rather than an outcome that has already been achieved.

Verizon has outlined an architecture that aims to bring mobile connectivity, broadband, cell towers, enterprise data centers, and residential connectivity onto a common high-capacity backbone. This converged approach is designed to deliver a seamless experience for consumers and businesses, connecting them through a single fiber optic network regardless of whether they are at home or on the move.

The optical fiber deployment under this agreement will provide the physical foundation for that converged infrastructure, supporting both consumer broadband expansion and enterprise-grade connectivity requirements.

AI Infrastructure and Data Center Connectivity

Beyond broadband, the Verizon and Corning fiber agreement addresses the growing demand for data center connectivity driven by artificial intelligence workloads. Under its AI Connect strategy, Verizon is deploying ultra-dense cables across major long-haul corridors to connect data centers. Heavy AI workloads require substantial throughput and low latency, making high-density fiber essential to support the scale of large language models and other compute-intensive applications.

The fiber optic network enabled by this supply agreement is intended to serve as both the broadband expansion backbone and the long-haul AI infrastructure corridor. This multi-use approach means the same physical fiber assets will support consumer internet access, business services, and the interconnection of AI data centers.

Among the optical solutions referenced in the agreement is Corning’s Contour Flow Cable, a flexible ribbon cable that allows more optical fiber to be deployed within existing conduit pathways. This technology is designed to maximize the capacity and bandwidth of Verizon’s infrastructure investments.

The Verizon and Corning fiber agreement is also structured to support Corning’s continued scaling of its manufacturing footprint in the United States. By securing long-term supply commitments, the deal provides Corning with the certainty needed to invest in production capacity for the fiber, cable, and connectivity solutions that Verizon requires.

Kyle Malady, CEO of Verizon Business, said the agreement allows the company to continue building a converged architecture at scale, uniting mobile and broadband connectivity while building the high-capacity backbone that AI infrastructure demands. Mike O’Day, Senior Vice President and General Manager of Corning Optical Communications, noted that the partnership reflects the intersection of broadband expansion and AI connectivity as part of the same opportunity, with Corning manufacturing its optical innovations at scale in the United States.

The AI for All Project is designed to go beyond general-purpose chatbots by creating AI agents that connect to actual public, medical, financial, education and housing services. Rather than users navigating multiple apps and websites, the AI finds the relevant service and carries them through each step. Services under the AI for All Project will roll out in stages, with the Kakao consortium planning a beta service in September and SK Telecom unveiling its beta in October before a full launch within the year.

SK Telecom, KT and Kakao Selected for AI for All

Each of the three consortiums brings a distinct approach to delivering nationwide AI services.

The SK Telecom consortium, which includes Goodoc, Shinhan Card and TMAP Mobility, has proposed an action-oriented daily-life AI assistant. The service would be accessible through phone calls and text messages, handling tasks ranging from government paperwork and health management to job hunting. By linking with Government24 and PASS, users could obtain 83 types of certificates. The assistant would also gather medical checkup results, explain them in plain language and guide users to nearby hospitals. For job seekers, it would track application deadlines and assist with cover letters and interview preparation. SK Telecom plans to build its service around its proprietary AI foundation model while also drawing on a range of domestic AI models.

KT’s consortium includes Upstage, Musinsa and Zigbang. KT has proposed a personalised AI service that ties together housing, education, finance and shopping into a single integrated assistant linked to commerce platforms. With user consent, the AI remembers interests, schedules and tasks in progress, and proactively suggests what needs to be done next. KT’s consortium will draw on multiple Korean AI models and will also deploy domestic AI infrastructure technologies, including Korean-made neural processing units.

The Kakao consortium, which includes LG Uplus and Seoul National University Hospital, plans to make KakaoTalk the main point of contact for AI services. Users would access a general-purpose AI chatbot within KakaoTalk and be connected to specialised AI agents covering medical, financial, tax, education, employment and care services. The service will run on in-house AI models including Kakao’s Kanana and K-EXAONE.

Nationwide AI Services Planned After Beta Testing

The government plans to sign agreements with the three selected operators and will provide them with 512 Nvidia B200 GPUs this year to support the AI for All Project. The nationwide service is planned to launch after the beta phase is completed.

A key priority of the AI for All Project is accessibility. SK Telecom is making its AI available through phone calls and text messages, while Kakao and LG Uplus plan to offer access through KakaoTalk and phone calls. KT will build a dedicated app and website while also applying AI functions to existing services. These channels are intended to ensure that older adults and people less comfortable with digital devices can use AI services through familiar tools.

The project also aims to create real-world use cases for the domestic AI industry. All three consortiums plan to use multiple Korean AI models, turning large-scale public services into a proving ground for domestically developed artificial intelligence technologies. The role of major telecom operators in delivering these nationwide AI services underscores the infrastructure capabilities required to support the AI for All Project at scale.

The collaboration was unveiled at NVIDIA’s AI Summit in San Jose, where Zayo confirmed it would leverage NVIDIA’s networking technologies alongside its own extensive fibre footprint to build out dedicated AI transport corridors. The effort responds to a rapidly growing need for network pathways capable of handling the enormous data flows generated by AI training and inference workloads.

Why New Fibre Corridors Are Essential for AI Infrastructure

The rise of large-scale AI factories has created unprecedented demand for network capacity between compute facilities. Traditional internet and enterprise traffic patterns are being reshaped by AI workloads, which require ultra-low-latency, high-throughput connections between clusters of graphics processing units spread across multiple sites. This AI network expansion addresses a bottleneck that threatens to constrain the buildout of AI infrastructure if connectivity does not keep pace with compute deployment.

Zayo operates one of the largest independently owned fibre networks in North America, spanning more than 141,000 route miles. The company said it is now directing significant capital toward expanding capacity on routes that serve the highest concentrations of AI data centres. These corridors connect key markets where hyperscale operators, cloud providers, and AI developers are clustering their facilities.

Dan Caruso, chief executive of Zayo, said the partnership with NVIDIA reflects the reality that networking is becoming as critical as compute in the AI supply chain. He noted that Zayo’s fibre assets position the company to deliver the physical infrastructure layer that AI factories depend upon, and that the collaboration will accelerate the pace at which new capacity reaches the market.

Technical Details of the Fibre Expansion

Under the initiative, Zayo plans to deploy additional dark fibre pairs and lit services across high-demand AI corridors. The company will integrate NVIDIA’s networking technology stack to optimise data transport between AI facilities. This includes support for back-end AI cluster networking, where massive volumes of data must move between GPUs during model training.

The AI network expansion also encompasses upgrades to existing fibre routes, increasing the density of available fibre strands on pathways that connect established data centre campuses. Zayo confirmed that the buildout would span multiple phases, with initial deployments targeting corridors in the western and central United States where AI facility construction is most concentrated.

Kevin Deierling, senior vice president at NVIDIA, said that scaling AI infrastructure requires a network fabric that extends beyond the walls of individual data centres. He described the collaboration with Zayo as an important step in ensuring that connectivity keeps pace with the rapid expansion of AI compute capacity across the continent.

The initiative also aligns with broader industry trends. Demand for fibre connectivity between data centres has accelerated sharply as enterprises and hyperscalers invest heavily in AI capabilities. Network operators across North America are racing to secure and light new fibre routes, making AI network expansion one of the most active segments of telecommunications infrastructure investment.

Next Steps in the Rollout

Zayo said the first phase of new fibre deployments under the collaboration is expected to commence in the second half of 2025, with additional corridors to follow through 2026. The company indicated it would announce specific route details and capacity milestones as construction progresses.

The partnership positions both companies to address what industry observers regard as a critical constraint on AI growth. As compute power continues to scale inside AI factories, the connecting network must evolve in parallel. This AI network expansion effort represents a direct response to that challenge, with Zayo committing substantial fibre resources and NVIDIA contributing its networking expertise to ensure the infrastructure is fit for purpose.

Further updates on corridor-level deployments and technical specifications are expected in the coming quarters as Zayo advances its buildout programme alongside NVIDIA.