The integration of Artificial Intelligence (AI) into the Radio Access Network (RAN) has transitioned from a theoretical research interest to a practical, albeit cautious, operational reality. At the recent Intelligent RAN Forum, a panel of industry leaders from major operators and infrastructure vendors converged on a consensus: AI is currently delivering significant value through operational efficiency, energy savings, and network optimization. However, the elusive "killer app" that would drive new, direct revenue streams remains missing.
As the telecommunications sector grapples with stagnant market growth, the debate surrounding AI-RAN has shifted from "if" to "how" and "where." The consensus among industry experts is that while AI-driven automation is successfully trimming operational expenditures (OPEX), the industry is still struggling to justify the capital investment required to transform the RAN into a revenue-generating platform for new, AI-centric enterprise services.
Main Facts: The Current State of Play
The primary takeaway from the forum is that AI is successfully performing the "heavy lifting" of network management. Experts identified three core areas where AI-driven RAN is already paying for itself:
- Energy Optimization: AI algorithms are dynamically managing power consumption across cell sites, turning off underutilized hardware during low-traffic periods without compromising coverage.
- Operational Automation: Through automated fault diagnosis and self-healing mechanisms, operators are reducing the time required to resolve "alarm storms"—the influx of error alerts—from days to mere minutes.
- Capacity Planning: Predictive analytics are enabling more surgical capital expenditure (CAPEX) deployment, allowing operators to delay hardware upgrades by squeezing maximum performance out of existing infrastructure.
Despite these clear operational wins, the panel—moderated by Tom Camp, executive director of the Telecom Council—highlighted a significant disconnect: the investment thesis for AI-RAN is currently split between proven, incremental cost savings and unproven, aspirational revenue growth.
Chronology: From Pilot Programs to Industrial Integration
The journey of AI-RAN over the past 18 to 24 months reflects a period of intense experimentation.
- 18 Months Ago: Vendors and operators began reporting breakthrough trial results. Notable data points included cell-edge uplink throughput improvements ranging from 20% to 50%. These figures served as the initial proof points that AI could fundamentally improve network performance.
- 12 Months Ago: The focus shifted toward integrating AI into existing network management tools. Operators like Turkcell began building proprietary, AI-driven optimization frameworks, moving beyond third-party off-the-shelf solutions.
- Present Day: The industry is now in a "pragmatic phase." The conversation has moved away from laboratory benchmarks toward the realities of legacy integration, skill gaps, and the difficulty of measuring "customer experience" improvements resulting from AI-enhanced network backends.
Supporting Data and Economic Realities
The economic argument for AI-RAN is complex because the benefits are distributed across multiple organizational silos. As Güneş Kesik, associate R&D director at Turkcell, pointed out, energy savings reduce OPEX, while improved capacity utilization delays CAPEX. Because the benefit is not captured in a single ledger, it is difficult for Chief Financial Officers (CFOs) to allocate a dedicated "AI-RAN budget."
The Performance vs. Maturity Gap
While technical performance in trials is high, large-scale commercial deployments remain scarce. Rob Hughes, head of wireless marketing at 1Finity (Fujitsu’s network infrastructure brand), noted that operator skepticism is rooted in maturity, not technology. "You see some trials and some smaller deployments, but not large-scale deployments yet," Hughes explained.
The Complexity Trade-off
There is a growing awareness that AI should not merely add another layer of technical debt. Kesik noted that the complexity removed through automation must significantly outweigh the complexity introduced by the AI models themselves—such as data integration, software lifecycle management, and compute overhead.
Official Responses and Strategic Perspectives
The Operator Perspective: Pragmatism First
Gabriel Ionita, director of group partnering and business development at Deutsche Telekom, emphasized that pilots must have a clear "customer-facing result." He expressed skepticism toward internal network improvements that do not eventually translate into a better experience for the end-user. For Ionita, the priority is clear: "Make better use of the network we already have."
Turkcell’s Güneş Kesik echoed this sentiment, suggesting that while the temptation to invest in new revenue-generating products is high, the reality of the market demands a focus on automation. "We are asked to be pragmatic," he said. The current strategy is to prioritize AI-for-the-network, with the expectation that the network-for-AI will follow only when market demand creates a clear path to profitability.
The Vendor Perspective: Building the Ecosystem
Rob Hughes advocated for an opportunity-led approach. By tying deployments to a specific, paying enterprise customer, operators can de-risk the investment. He noted that hyperscalers (cloud providers) are dominant in AI cloud infrastructure, but they lack the physical advantages that telcos possess: site real estate, reliable power, and deep-seated enterprise relationships.
"Your traditional business is strengthened if you can add more value by moving up the value chain to the customer," Hughes argued. He suggested that AI-RAN, when combined with private wireless networks, provides a unique opportunity to embed connectivity deeper into the core operations of enterprise clients.
Implications: The Looming Competitive Threat
The most significant implication of the forum’s discussion is the "circular dilemma" of the edge. Application developers are hesitant to build specialized RAN-based applications because the infrastructure (the edge) is not yet ubiquitous. Simultaneously, operators are reluctant to deploy massive edge infrastructure until there is a steady stream of applications to justify the cost.
The Hyperscaler Shadow
The industry faces a ticking clock. As operators deliberate over the business case for AI-RAN, hyperscalers are aggressively moving into the edge space. The panel raised a critical question: Can operators afford to wait for demand to materialize, or will they cede their valuable real estate and connectivity advantages to cloud providers who are already moving to capture the edge market?
The Path Forward
To bridge the gap between efficiency and revenue, the panel suggested three criteria for "investable innovation":
- Measurable Economic Impact: Moving beyond technical KPIs to show clear financial gain.
- Repeatability: Ensuring that AI models survive vendor updates, topology changes, and the integration of new spectrum bands.
- Scalability: Moving from isolated, small-scale trials to standardized deployments that can be replicated across thousands of sites.
Ultimately, the consensus at the Intelligent RAN Forum was one of cautious optimism. The industry has proven that AI is a powerful tool for cost reduction and network efficiency. However, the transformation of the RAN into a robust engine for revenue growth remains an ongoing, complex evolution. Operators that successfully blend their existing physical assets with the agility of AI-driven, software-defined networks will likely define the next decade of telecommunications, provided they can solve the "demand" puzzle before the competitive landscape shifts further in favor of cloud-native rivals.
As the industry looks toward 2030, the pressure to demonstrate not just technical superiority, but true economic value, will only intensify. For now, the "Ops First, Revenue Later" strategy appears to be the only path that balances immediate financial reality with long-term strategic ambition.
