The rules governing the ownership, valuation, management, and allocation of digital infrastructure have changed dramatically in recent years, and the rate of change is accelerating. In 2025, the value of global digital infrastructure M&A deals increased by more than 60% over the previous year. What were once relatively straightforward, reliably income-producing holdings and technologies have evolved into a far more various and complex assortment of assets and services. Today, the challenge for operators and investors lies in gaining a deeper understanding of where value will accrue.
In this environment, broad exposure to digital assets raises the possibility of riskier outcomes because value creation depends on a wider range of asset-level portfolio choices. For investors, that means adopting a due diligence approach, grounded in fundamentals, to hunt in misclassified seams. For operators and infrastructure-linked businesses—including telcos, tower and fiber platforms, data center operators, engineering, procurement, and construction firms (EPCs), power-adjacent service providers, and select OEMs—it means decomposing portfolios and matching each asset, capability, or service offering to the right growth, partnership, separation, or repurposing path.
If investors and operators are to gain from growth in this swiftly expanding asset class, they must understand the reasons behind the acceleration and the technology trends generating the increased interest, and must adapt their underwriting and strategic frameworks accordingly.
Digital Infrastructure Was Once a Coherent and Neatly Defined Category
Communications assets formed the original foundation of digital infrastructure. Assets such as zoned cell tower sites, exclusive building access rights, and constrained rights-of-way tended to satisfy very clear economic criteria: near-zero churn and high cash-flow visibility through predictable, long-term contracts. The key variable was the high level of control that owners exerted without much difficulty over these assets.
Fiber infrastructure played a more varied role, introducing greater heterogeneity into what companies otherwise treated as a coherent infrastructure category. Capturing the value of fiber required active commercial execution rather than simple asset control. Certain long-haul routes and metro rings benefited from route scarcity and anchor tenants, but many access and fiber-to-the-home (FTTH) networks depended heavily on take-up rates, marketing spending, and ongoing operational and commercial execution.
Over time, the digital infrastructure category expanded dramatically. Other types of data centers, edge facilities, power-linked assets, chip producers, services and software companies, and more distant asset classes provided a broader set of digital and AI-adjacent businesses that have since been drawn into the same investment universe.
New Boundary Conditions
The ongoing expansion of digital infrastructure has significantly enlarged both the opportunity set and the control challenges for operators and investors, increasing sensitivity to utilization, customer mix, supply chain constraints, and reinvestment cycles. The need for AI infrastructure has only accelerated this trend. Computing-power-intensive data halls, power interconnection assets, cooling systems, and AI-specific retrofits now fall into the same broad category as traditional infrastructure. As a result, portfolios continue to diversify as the boundary of digital infrastructure expands to include AI infrastructure, as our 2026 Infrastructure Investment report shows. (See Exhibit 1.)
The costs of misclassification have also become more visible. Assets that were grouped together under a common infrastructure label in the past are increasingly exhibiting diverging economic behavior. Some, like cell towers, continue to support long-duration, infrastructure-style underwriting. Others—such as FTTH, newer satellite constellations, and the conversion of telecom central offices into data centers—depend far more on operational execution, customer mix, technology trends, utilization, and reinvestment. When companies overlook those differences, they often hold, value, and sell assets under assumptions that no longer match their actual performance.
This heightens the challenge for investors looking to deepen their exposure to digital, which entails understanding more deeply the technology trends, commercial risks, and variables that can form infrastructure-like properties. The gap shows up most clearly in transaction outcomes. Capital remains abundant for scaled, well-understood assets, especially for core data centers and hyperscale-adjacent platforms. Execution-heavy and transitional assets such as B2C fiber and cable, however, have faced wider valuation gaps, longer exit timelines, and a narrower buyer universe as investors reassess churn, terminal penetration targets, capital intensity, and operating complexity.
Market data highlights this divergence. According to a recent survey by TMT Finance, a data intelligence company, approximately 83% of respondents expect M&A activity in digital infrastructure to increase, reflecting a heightened investor appetite. But deal flow is increasingly concentrating in data centers and hyperscale-adjacent transactions. In the US, for example, data center transaction value grew more than threefold—from roughly $20 billion to over $63 billion—from 2024 to 2025, according to S&P Global Market Intelligence.
Meanwhile, many execution-heavy and transitional assets face measurably slower exit timelines and wider valuation gaps. The average infrastructure asset holding period stretched to 7.6 years in 2025, up from 6.1 years in 2021, according to our analysis of the top 50 infrastructure general partners. Concerns about valuation gaps have moved in the same direction: in Europe, the share of surveyed investors who cited valuation gaps as the primary barrier to deal completion jumped from 48% to 78% year-on-year; and in the US, 63% of respondents cited high valuations as the top obstacle to dealmaking.
The Strategic Imperative
In this environment, the strategic focus is moving away from broad asset classes and simple distinctions. Risk, upside, and repurposing potential now vary at the asset level rather than the organizational level. (See Exhibit 2.) As a result, portfolio strategy is shifting from structural separation to asset-specific allocation, with explicit decisions about which assets to invest in or grow and which to exit or reposition.
Investors must increasingly rely on tangible indicators such as churn rates, renewal behavior, repricing frequency, and capital intensity to understand how specific assets perform in good times and, more importantly, under stress. These signals help determine whether it is best to own a particular asset as a long-duration yield asset, an execution-heavy growth business, or a candidate for repurposing. As measured by global infrastructure deals from 2020 to 2025, 79% of enterprise value growth came from revenue growth, ahead of margin or multiple expansion. The investors generating the strongest returns are those that improve the operating performance of the underlying business, rather than depending on rerating valuations.
For operators, this means that portfolio strategy can no longer rely solely on single structural frameworks, such as the distinction between NetCo and ServCo. Instead, they need to decompose, repurpose, partner, and selectively grow mixed portfolios on the basis of asset-specific economics. Operators should supplement these strategies with attention to the opportunity posed by rapidly growing interest in AI, with regard to both the fundamentals of their business and the potential appreciation in public stock markets when exposed to AI’s momentum.
Investors Should Hunt in the Seams
For investors, this shift means that emphasizing broad exposure to crowded areas won’t maximize outcomes. Value increasingly concentrates in specific asset classes where quality exists, but ownership structures, operating models, or market narratives remain misaligned. Opportunities for investors to evaluate include the following seven:
- Multidwelling Unit (MDU/Bulk) Fiber and Embedded Access Networks. The value driver in this area is control over building access, portfolio-wide exclusivity, and long-term bulk contracts, rather than retail subscriber growth. These assets often exhibit lower churn than open-access FTTH networks, but they remain underaggregated due to operational complexity. As such, they are poised for growth through improved commercial execution.
- Specialized Data Center Services. As ownership of data centers becomes crowded and capital intensive, investment in capabilities that are critical to data center build, deliverability, and maintenance cycles—such as construction services, networking, commissioning, cooling, power integration, and life-cycle operations—can capture AI-driven growth without assuming concentrated risk. This opportunity also applies to emerging supply chains and capabilities in support of orbital data centers in space. Demand for these workforces and specializations is increasing rapidly, so companies that are available for adjacent plays need to quickly scale capabilities through talent acquisition and M&A.
- Transitional and Stranded Network Assets. Although communications assets such as copper networks, subscale hybrid fiber coaxial networks, satellite ground infrastructure, and legacy central office facilities may no longer fit growth-oriented portfolios, they can still generate value through harvest, decommissioning, migration, or repurposing into shared infrastructure. Investors often discount these assets because they sit awkwardly within portfolios built for growth, but that same mismatch can make them attractive to buyers with a different time horizon, cost structure, or operational thesis toward higher shared utilization.
- Indoor Wireless and Controlled Venues. The value of these assets depends on venue quality, venue willingness to fund, carrier utilization, and carrier contract durability. Value creation depends on separating long-lived infrastructure from project-based deployments or services across private networks, Wi-Fi, and other growth areas, and potentially separating core co-located facilities in Tier 1 venues from other business lines, all of which may present opportunities for consolidation and leaner operating models.
- Utility-Adjacent and Embedded Communications. Cloud- and software-based communications workflows and software embedded in utilities, logistics, and telecom operations are single-threaded—built around a single, deeply embedded communications stack—with strong contractual behavior and high switching costs, along with regulatory and mission-critical characteristics, presenting infrastructure-like durability.
- Physical AI and Robotics Infrastructure. Emerging as a key adjacency as AI begins to act in the physical world, common operating layers are taking shape across robotics and industrial automation. These include edge inference, sensors, robot operating systems, and fleet management infrastructure. When shared across use cases from warehouse automation and Industrial Internet of Things (IIoT), these layers could deliver the platform economics and utilization characteristics that attract infrastructure investors.
- Fiber for AI Clusters and Computing Power (Neofiber). This is the next generation of high-capacity, physically diverse fiber infrastructure built to connect AI and hyperscale data centers located in new power and land-available markets. It addresses the growing mismatch between where computing power infrastructure can be built and where sufficient network capacity, route diversity, and rights of way already exist. The opportunity for investors lies in funding and aggregating scarce rights of way, regional corridors, civil work needs, and access infrastructure through anchor-backed platforms that connect emerging AI markets to existing networks and de-risk data center campus delivery.
Four Key Steps for Operators
For operators, boundary blur is no longer just a classification issue. As balance sheet pressures and growth mandates intensify, the stakes are high: BCG's 2026 Telco Value Creators Report found that asset-heavy NetCos spun out from operating businesses delivered a five-year total shareholder return of –4%, the weakest of any telco archetype. For their part, 5G and fiber-to-the-home rollouts have yet to meaningfully lift average revenue per user despite substantial capital outlay.
To convert this pressure into exposure to AI tailwinds, operators must identify and execute creative opportunities to reallocate capital and reassess their portfolio management strategies. Many operators manage portfolios that contain not just physical assets but also a range of services and resources that exhibit different economic behaviors and provide unique opportunities for creating value. For example, some may be able to build and manage data centers and leverage their distributed facilities, their dense fiber, their enterprise relationships, and even their trusted billing platforms, either making these available as a service or spinning them off into separate offerings. (See Exhibit 3.)
In light of these options, the task for capital allocation and portfolio management is no longer simply to decide whether to hold or sell. It is to recompose the portfolio asset by asset and service by service, matching each hard or soft asset to the ownership model, capital strategy, and value creation path.
To that end, operators need to take four key steps:
- Decompose the portfolio at the asset and service level. Legacy business-unit structures can obscure the underlying risk, capital intensity, and value-creation potential of individual assets and services. Operators should distinguish between assets that behave like long-duration infrastructure and require active operational execution; transitional assets that may be strategically expendable, such as land, facilities, towers, copper networks, and central offices; and capabilities such as data center development and operations or fiber and network management that may create greater value when made available to third parties.
- Reposition ownership; don’t simply relabel assets. Partnerships, joint ventures, minority investment, and carve-outs can align assets with the appropriate capital base, investment horizon, and operating model. This increasingly includes structures designed to support sovereign AI and the developing systems and governance arrangements that determine where AI computing power and data reside, who can access them, and how security and regulatory requirements are enforced. Governments around the world are seeking domestically located and controlled AI training and computing power capacity, often through joint ventures or minority investments involving local operators that can meet security, data-residency, and regulatory requirements.
- Repurpose existing assets and capabilities while funding selective growth. Operators can capture AI-driven upside by redeploying owned real estate, central offices, conduits, power-adjacent sites, fiber infrastructure, and service capabilities to support data centers and high-capacity network build-outs. However, this growth strategy is credible only when paired with clear decisions about which legacy networks, facilities, and businesses the operator should retain, repurpose, separate, or exit.
- Reinvent internal operations around AI-first principles. Many operators already possess much of the infrastructure, technical talent, data, and partner ecosystem required to deploy AI at scale internally. The harder task involves determining where AI can create the most value, redesigning processes around those use cases, and aligning technology, governance, talent, and investment behind a coherent transformation agenda. Done well, this step can improve network planning, service assurance, customer operations, field service, and corporate functions, while demonstrating to customers and investors that the operator can translate AI capabilities into measurable business outcomes.
The rapid blurring of the boundaries of digital infrastructure has generated a much expanded range of options for investors and operators alike. What was once a relatively straightforward asset class has diversified into a complex, heterogeneous mix of hard assets, services, and resources—a mix increasingly dominated by the rapid rise of assets, technologies, and services dedicated to supporting the AI revolution.
While this has increased the challenges involved, it has also given rise to a wealth of new opportunities for investors and operators. Taking advantage of these possibilities will require a data-driven, market-aware process that focuses on expert portfolio management and pays constant attention to the ramifications of AI’s impact.
