The Connectivity Bottleneck: Why AI Growth Now Depends on Telecom Infrastructure
Artificial intelligence is often discussed as a compute story: chips, data centers, power, cooling, and model performance. But the next constraint is increasingly a connectivity story. AI workloads require vast movement of data across data centers, cloud regions, enterprises, devices, edge locations, and end users. Without high-capacity, low-latency, resilient telecommunications infrastructure, the value of compute cannot be fully realized.
For telecommunications CEOs, this creates both an opportunity and a risk. PwC describes AI and data centers as a new infrastructure super-cycle for telecom, while warning that traffic growth is rising faster than revenue. Deloitte similarly notes that AI, data services, and network APIs remain small in 2026 but could become material over time. The question is whether operators can convert AI-driven demand into profitable infrastructure economics rather than simply carrying more traffic on already capital-intensive networks.
AI growth will reward telecom operators that can turn connectivity demand into disciplined infrastructure value: prioritized fiber, intelligent network design, enterprise-grade reliability, commercial segmentation, and capex governance managed as one operating system.
AI Demand Is Moving from Compute Capacity to Network Capacity
The AI infrastructure buildout is changing how networks are used. Training clusters, inference workloads, enterprise AI applications, autonomous agents, connected devices, sovereign AI environments, and edge computing all depend on fast, reliable movement of data. The network is no longer only a downstream utility. It is becoming part of the performance architecture of AI itself.
This matters because AI traffic is not evenly distributed. Some traffic will concentrate around hyperscale data centers and cloud regions. Some will move to enterprise campuses, industrial sites, healthcare environments, financial institutions, public-sector workloads, and edge locations. Some will require deterministic performance, predictable latency, secure routing, or service-level guarantees that commodity broadband cannot provide.
For telecom operators, the growth question is therefore more precise than total traffic volume. Leaders need to know where traffic is valuable, where it is only costly, which customers require premium performance, which routes and markets justify fiber or edge investment, and which use cases can be monetized through enterprise solutions, APIs, private networks, or managed services.
Traffic Growth Alone Does Not Create Telecom Value
Telecommunications has lived with a structural problem for years: usage rises faster than revenue. PwC projects global telecom service revenue to rise from US$1.15 trillion in 2024 to roughly US$1.32 trillion in 2029, a modest CAGR of about 2.8%, while mobile ARPU is expected to decline slightly over that period. That means operators cannot assume that AI-driven demand will automatically improve margins.
The risk is that telecom companies become the low-return infrastructure layer beneath higher-margin AI ecosystems. Hyperscalers, cloud platforms, device companies, application providers, and enterprise integrators may capture the majority of value while carriers carry the traffic, absorb reliability expectations, and fund network upgrades.
The strategic challenge is to identify where telecom infrastructure has pricing power. Low-latency enterprise connectivity, data-center interconnect, sovereign network services, secure connectivity for regulated workloads, edge-enabled applications, network APIs, and managed connectivity for mission-critical environments may carry different economics than mass-market data growth. The operator must be able to distinguish value traffic from volume traffic.
Fiber and Interconnect Are Becoming Strategic Assets
AI has intensified the importance of fiber and optical connectivity. Reuters reported in June 2026 that Amazon and Corning signed a multi-billion-dollar agreement to boost U.S. production of optical fiber and connectivity products used in data centers. Corning also said its optical fiber products are crucial to moving data between thousands of processors in AI data centers and that it had announced plans to expand U.S. optical connectivity and fiber production capacity.
Those investments reinforce a larger point: connectivity is becoming a strategic input to AI infrastructure. Telecom operators that own valuable routes, metro fiber, long-haul assets, enterprise access networks, subsea or regional connectivity, and data-center interconnect can participate in the AI buildout if they manage those assets with discipline.
The operational question is where to build, reinforce, lease, partner, or avoid. Not every route deserves capacity expansion. Not every market will generate premium enterprise demand. Not every data-center opportunity will create attractive returns once construction, maintenance, energy, rights-of-way, installation capacity, and customer concentration risk are considered. The winners will be those that apply investment logic with precision.
AI-Era Connectivity Requires More Than Speed
The AI-era network is not defined only by bandwidth. Enterprises increasingly care about latency, jitter, uptime, security, routing resilience, burst capacity, data sovereignty, and integration across fixed, wireless, cloud, and edge environments. Deloitte notes that operators may have an opportunity to integrate fixed and wireless networks, edge computing, security, and mission-critical workloads as enterprises shift toward integrated, outcome-based solutions.
That opportunity will require operators to move beyond generic connectivity products. A manufacturer running AI-enabled quality systems, a hospital using AI-supported imaging workflows, a financial institution supporting low-latency analytics, or a logistics provider coordinating connected assets may need a different network promise than a consumer streaming video.
Commercially, this means better segmentation. Operationally, it means network engineering, product management, sales, service assurance, field operations, and customer success must operate from the same service model. Premium performance cannot be sold by the commercial team if the operating organization cannot design, provision, monitor, and recover the service reliably.
Wireless, Fiber, Edge, and Satellite Must Be Managed as One Portfolio
AI-era connectivity will not be solved by one network technology. Fiber will be central to data-center interconnect, backhaul, enterprise campuses, and high-density markets. 5G and 5G standalone architectures will support mobility, private networks, fixed wireless access, and industrial use cases. Edge environments may reduce latency and support localized inference. Satellite and non-terrestrial networks can extend coverage and resilience in hard-to-reach areas.
Reuters reported that AT&T outlined a US$250 billion U.S. infrastructure investment plan over five years, focused on fiber broadband, 5G home internet, and satellite connectivity amid rising demand from AI, cloud computing, and connected devices. That type of portfolio approach reflects the market reality: coverage, capacity, cost, density, customer mix, and service promise vary significantly by market.
For telecom leaders, the question is not which technology wins universally. It is which access model creates the best economic and service answer in each market and customer segment. That requires market-level build economics, disciplined installation capacity, contractor governance, customer-density analysis, churn assumptions, service-cost visibility, and clear decision rules for when to deploy each technology.
Network APIs Can Help Monetize Infrastructure, but Only with Commercial Discipline
Network APIs are one pathway for operators to convert infrastructure capabilities into programmable services. GSMA Intelligence reported that 73 operator groups representing 285 networks and almost 80% of mobile subscribers worldwide were committed to GSMA Open Gateway as of H1 2025, with focus moving toward commercial availability and monetization.
The relevance to AI is direct. Developers and enterprises may need access to capabilities such as quality-on-demand, device location, edge compute, identity verification, fraud prevention, and service assurance. But API availability does not create revenue by itself. Monetization requires developer onboarding, partner channels, pricing, product management, support, SLA governance, and clear accountability for adoption.
Telecom operators should avoid treating network APIs as a side project owned only by technology teams. If APIs are to become a meaningful commercial layer, they must be governed like products: defined customer problems, sales motions, usage economics, billing systems, reliability standards, and partner operating routines.
Capex Discipline Becomes the Strategic Control Point
The AI buildout creates pressure to invest, but capital intensity is already one of the defining constraints of telecommunications. The risk is overbuilding capacity where operators cannot earn an adequate return or underinvesting in assets that become critical to enterprise and AI infrastructure demand. Both errors are expensive.
Capex discipline must therefore become a live management capability, not only an annual planning exercise. Leaders need to connect market intelligence, customer demand, network utilization, route-level economics, installation capacity, competitive positioning, pricing strategy, and capital governance. The decision should not be simply whether to expand the network. It should be where expansion changes the return profile of the business.
That requires stronger visibility across engineering, finance, enterprise sales, network operations, procurement, real estate, construction, and field service. When those functions operate in sequence rather than as one system, build plans slip, costs rise, capacity is misallocated, and leadership loses the ability to steer capital toward the highest-value demand.
The Brooks International Perspective
From Brooks International’s perspective, the connectivity bottleneck is an operating model challenge. AI may create the demand signal, but telecom operators will only capture value if they can translate that signal into disciplined network investment, service design, commercial execution, and operational reliability.
The highest-value opportunities often sit between functions: network planning and enterprise sales, product management and service assurance, capital allocation and field execution, engineering and customer success, procurement and construction, network operations and cybersecurity. If those interfaces are weak, the operator can spend heavily without improving return.
Brooks International helps leadership teams connect infrastructure strategy to measurable operating performance. In telecommunications, that means creating the cadence, accountability, and performance visibility required to prioritize markets, manage capital, govern build execution, control service quality, and convert network demand into enterprise value.
Brooks International’s view is that AI-era connectivity must be managed as a business system, not a traffic forecast. The companies that win will know where the network creates premium value, where traffic is dilutive, and how to align build, operations, and commercial execution around the most attractive opportunities.
What Telecommunications Leaders Should Be Asking Now
The leadership agenda should focus on whether AI-driven connectivity demand is being translated into profitable infrastructure value, not simply higher traffic volume.
• Which AI-era connectivity use cases create premium economics, and which simply add traffic without improving margin?
• Do capital plans distinguish between high-value routes, strategic data-center interconnect, enterprise demand clusters, and low-return capacity expansion?
• Can leadership see utilization, service quality, build cost, installation capacity, and commercial opportunity in one operating view?
• Are fiber, wireless, edge, fixed wireless, and satellite decisions being made through market-level economics rather than technology preference?
• Does the enterprise sales organization have products, pricing, SLAs, and delivery capability that match AI-era customer requirements?
• Are network APIs, edge services, and managed connectivity governed as commercial products with owners, economics, and adoption targets?
The Leadership Imperative
The AI infrastructure cycle will not reward every telecom operator equally. The market will reward operators that know where connectivity has strategic value and can deploy capital with precision.
The winners will not be those that simply carry more traffic. They will be the organizations that align network assets, enterprise demand, service assurance, pricing, and field execution around the use cases that matter most.
For telecommunications leaders, the mandate is clear: make connectivity a controllable value engine for the AI economy, not a low-return cost layer beneath it.
Related Insights