Login
Register
Sponsored
AI is now being adopted by communication service providers around the world. Amol Phadke, Chief Transformation Officer at Tech Mahindra, talks to TM Forum about the company’s future-looking focus and how it is helping CSPs in transforming themselves into intelligent digital infrastructure players by leveraging AI.
Amol Phadke
Sponsored by:
Reimagining the role of CSPs in the AI era
Amol Phadke is the Chief Transformation Officer at Tech Mahindra, which is a Platinum sponsor of DTW Ignite 2026. He outlines Tech Mahindra’s approach to AI transformation, advises what steps CSPs should be taking next, and explains the difference between AI-enabled and AI-native networks.
AP: Tech Mahindra’s approach to AI is evolving from experimentation to AI industrialization at scale—engineered for telecom outcomes. We are enabling CSPs to transform into AI-native enterprises by embedding intelligence across networks, IT, and business layers through a platform-led transformation.
Our differentiation lies in combining deep telecom expertise with AI engineering and purpose-built platforms such as the Orion platform, which accelerates the shift toward autonomous, zero-touch operations and scales AI across other domains within enterprise.
Our focus today is on helping CSPs move from pilot to production—leveraging reusable accelerators, telecom-specific AI assets, and integrated modernization approaches to deliver measurable outcomes in cost efficiency, customer experience, and new revenue streams.
Looking ahead, the evolution is toward self-optimizing, intent-driven networks and enterprises, where AI agents collaborate across domains—network, IT, and customer—to proactively drive outcomes such as zero-touch operations, hyper-personalized services, and faster time-to-market.
Our AI Foundry framework is focused on addressing all aspects of the CSP ecosystem, including data, infrastructure, and models.
AP: During DTW Ignite, we are inviting visitors to see how leading CSPs can bring the AI‑native telco to life through a future-forward playbook built on three pillars: Grow. Empower. Transform.
Indeed, at Tech Mahindra, we’re not only describing the future—we’re already engineering it at scale. Our holistic, end-to-end solutions cover all aspects of autonomous networks, from planning and build through to provisioning and operations.
By combining digital twins, agentic AI, and domain-specific models, we enable intent-driven, closed-loop autonomous operations, taking CSPs from manual workflows to predictive, self-healing networks.
With an Autonomous Network Operations Platform (ANOP), we are delivering live, closed-loop autonomous operations, not fragmented use cases. Identification of the high-value use cases for autonomous networks is key for any operator. Building a layered ecosystem to deliver the autonomous networks use cases with a single data platform and using fine-tuned models is delivering results to our customers.
Our AI platform for network autonomous operations is proven to help CSPs achieve Level 3.0 – 3.5 network autonomy. For example, at a tier 1 European telco, we have collaborated with NVIDIA to develop a domain-specific language model (DSLM), which is being adopted in network field operations, and this reusable asset is being extended for adoption at other CSPs.
Building on this, Tech Mahindra’s approach to AI transformation goes beyond deploying individual use cases—we have introduced a new execution framework designed for AI-driven enterprises. This framework comprises three integrated constructs: structured human + AI agent teams (Vector Squads), productized outcome units (Service Tokens), and a novel analytical lens for pricing decisions (the Pricing Model Suitability Quadrant).
Vector Squads enable predictable, auditable, and scalable execution patterns, making true outcome-based pricing operationally feasible. Service Tokens further productize outcomes into measurable, contractible units—decoupling cost from volume and enabling cloud-like SKU catalogs for enterprise services.
Together, this model addresses long-standing industry challenges around variability, governance, and attribution, positioning outcome-based pricing in the “sweet spot” of high value and high measurability. It enables faster cycle times, supports stable margins, and drives enterprise-wide applicability—offering a practical blueprint for scaling AI-driven operating models.
AP: Deploying AI on the network is a good starting point—but it is no longer sufficient. The real transformation happens when AI is deployed within the network itself.
When AI sits on top of the network, it primarily delivers observability, analytics, and post-event optimization. It helps us understand what happened and, increasingly, predict what might happen next. But telecom networks today operate at a scale and speed where decisions must be made in milliseconds—especially in domains such as RAN scheduling and real-time traffic management.
This is where AI within the network becomes critical, and embedding AI inside the network fabric fundamentally changes how the network functions.
For instance, it enables real-time, closed-loop decision-making, where sensing, analysis, and action happen inline, not as a separate step. It supports self-optimizing and self-healing behaviors, allowing the network to adapt dynamically to changing conditions.
It also transforms the network into an AI-native system, where AI is embedded across RAN, core, and operations layers—not added as an external capability.
In other words, AI on the network improves efficiency, while AI within the network enables autonomy, and this distinction is critical.
And autonomy is non-negotiable as we move toward 5G-Advanced and 6G, which creates the foundation for the future. Intelligence must be deeply integrated into the control and data planes, orchestrating resources, policies, and services in real time.
There is also a strategic dimension here. When AI is embedded within the network, the network itself becomes a programmable intelligence layer. Capabilities such as QoS, location, and compute can be exposed via APIs, and CSPs can move from delivering connectivity to offering AI-driven, intent-based services and platforms.
At Tech Mahindra, we view this as the shift from AI-enabled networks to AI-native networks—a transition where every part of the lifecycle, from design to operations, is AI-driven and continuously evolving.
Ultimately, if the ambition is to build truly autonomous, monetizable, and future-ready networks, AI cannot remain outside the system. It must become part of the network’s DNA.
AP: The shift beyond connectivity is not just a technology transition—it is a business model reinvention. CSPs will move faster when they treat the network not as a product, but as a platform for intelligence and value creation.
At its core, this is the leap from moving bits to moving intelligence—where connectivity becomes the distribution layer for AI-driven services. This shift accelerates when AI agents, models, and inference engines operate at the edge, in real time, across billions of endpoints.
To realize this, CSPs must move decisively on various fronts. For instance, they need to productize intelligence and turn network data, AI models, and real-time insights into consumable services and APIs, not just internal capabilities.
Other requirements are to build an intelligence stack, extend beyond connectivity into a layered architecture that integrates data, AI models, orchestration, and exposure capabilities. They also should push sensing and inference to the edge, enabling real-time decision-making by embedding distributed sensing and edge AI, supported by sovereign and domain-specific models.
In summary, this is not an incremental evolution, it is a structural shift where value is created. This is because the future isn’t about faster connectivity; it’s about who owns, orchestrates, and monetizes the intelligence layer on top of it. And that is the race we are helping our customers win—faster and at scale.
AP: If I distil this down, these are the use cases that move CSPs from reactive operations to autonomous, AI-driven execution across the full lifecycle.
AI systems that predict, diagnose, and resolve issues autonomously, moving away from ticket-based operations to zero-touch operations, bring significant impact. Predictive and intent-driven maintenance, AI-led closed-loop network optimization, and agentic NOC which reason, correlate, and act across domains are other high-value use cases.
The real impact comes when AI is no longer applied as point solutions, but as a horizontal layer across the entire network lifecycle—resetting operations from manual and reactive to autonomous, intent-driven, and continuously optimized.
