MTN’s move to Azure for its Enterprise Value Analytics platform marks a watershed moment for telco cloud adoption in Africa: the operator says EVA 3.0 has been re‑engineered on Microsoft Azure (using Azure Databricks and Microsoft security tooling) to deliver faster analytics, earlier operational visibility and the scale to ingest extremely large daily volumes of telemetry — a blueprint MTN is positioning as a continent‑wide model for data‑driven telco transformation.
The telecommunications industry has been shifting quickly from siloed, on‑premises analytics stacks to unified, cloud‑native data platforms that can power real‑time operations, personalization and AI. Microsoft and other hyperscalers have framed this shift around two connected promises: scale for massive telemetry and integrated security and governance so operators can use sensitive customer and network data responsibly. Microsoft’s industry messaging — including a telco data model inside Microsoft Fabric and numerous telco partnerships announced in 2024–25 — makes clear that cloud + AI is now the dominant vendor narrative for network operators modernizing on a platform approach. MTN’s programme sits squarely inside that industry movement. The operator launched a multi‑year cloud strategy (Project Nephos), created a Cloud Centre of Excellence to drive capability across its operating companies, and has pursued large‑scale skills certification with Microsoft’s Enterprise Skills Initiative. MTN’s own newsroom reports a company‑wide certification milestone — more than 1,250 staff certified on Microsoft Azure frameworks in 2025 — underscoring that the company is investing in the people and processes needed to run a cloud transformation at scale.
At the same time, the public emphasis on certifications and workforce training highlights another imperative: cloud modernization must be matched by a local skills pipeline. MTN’s reported certification volume suggests an emphasis on building talent internally — a critical signal for governments and training bodies to scale cloud and security education across the region.
For African telcos, MTN’s EVA 3.0 represents both an operational template and a cautionary tale: cloud platforms can unlock dramatic new capabilities, but they must be paired with robust governance, security, cost controls and people capability to deliver sustainable value.
Source: innovation-village.com MTN, Microsoft Launches Africa’s Largest Telco Cloud Platform - Innovation Village | Technology, Product Reviews, Business
Background
The telecommunications industry has been shifting quickly from siloed, on‑premises analytics stacks to unified, cloud‑native data platforms that can power real‑time operations, personalization and AI. Microsoft and other hyperscalers have framed this shift around two connected promises: scale for massive telemetry and integrated security and governance so operators can use sensitive customer and network data responsibly. Microsoft’s industry messaging — including a telco data model inside Microsoft Fabric and numerous telco partnerships announced in 2024–25 — makes clear that cloud + AI is now the dominant vendor narrative for network operators modernizing on a platform approach. MTN’s programme sits squarely inside that industry movement. The operator launched a multi‑year cloud strategy (Project Nephos), created a Cloud Centre of Excellence to drive capability across its operating companies, and has pursued large‑scale skills certification with Microsoft’s Enterprise Skills Initiative. MTN’s own newsroom reports a company‑wide certification milestone — more than 1,250 staff certified on Microsoft Azure frameworks in 2025 — underscoring that the company is investing in the people and processes needed to run a cloud transformation at scale. What MTN says EVA 3.0 delivers
The publicly circulated account of the EVA 3.0 rollout (announced in coverage of the MTN–Microsoft collaboration) highlights several core advances:- A migration of Enterprise Value Analytics (EVA) to Microsoft Azure, where EVA 3.0 is reported to run on Azure Databricks and be secured with Microsoft Defender.
- Massive ingestion and processing scale: the published report states EVA 3.0 processes roughly 22 billion records per day, handles 1,700+ data feeds, and executes 800+ analytic workflows.
- Faster time‑to‑insight and real‑time analytics, giving MTN earlier visibility into service performance, faster incident response, and better personalization or offer design.
- A platform designed to support responsible AI and governed analytics across the group, and to serve as a replicable blueprint for MTN’s other markets across Africa.
Why Azure + Databricks is a logical foundation for a telco analytics backbone
Scale and unified analytics
Azure Databricks is engineered around Apache Spark and the lakehouse paradigm, combining streaming and batch processing with ML lifecycle tooling. For telcos that must correlate streaming network telemetry with billing, CRM and OSS/BSS records, a single lakehouse that supports high‑throughput ingestion and parallel analytic workflows reduces complexity versus dozens of bespoke point solutions. Databricks on Azure continues to receive platform investments and partnership expansions that improve integration with Microsoft Fabric and other Azure services, making it a natural choice for enterprise telco workloads.Security and governance
Large telco datasets are sensitive: customer personal data, location and network traces require careful access controls and strong operational security. Microsoft’s Defender portfolio (Defender for Cloud and related XDR services) plus Azure identity services and data governance tooling are designed to provide continuous posture assessment, workload protection and entitlements control across cloud resources. When operators pair Databricks with Defender and Microsoft’s identity and governance stack they gain a path toward automated detection, policy enforcement and compliance reporting at scale — an essential requirement for multi‑country telcos subject to varied data‑sovereignty and telecom regulation.Telco‑specific practice and data models
Microsoft’s recent investment in a telco industry data model inside Microsoft Fabric aims to shorten time to value for operators by offering telecom‑specific schemas and prebuilt integration patterns. That work complements a Databricks lakehouse by providing the schemas and semantics that make telemetry usable for cross‑domain AI and insights. MTN’s stated intent to make EVA 3.0 a blueprint for other markets aligns with this pattern: reuse the platform stack and embed a telco data model to accelerate repeatable deployments.Concrete benefits MTN will likely see (and already claims)
- Faster incident detection and response. Real‑time analytics and anomaly detection across 22B daily records (reported) should provide earlier warning of network degradation and service incidents — lowering mean time to repair.
- Better operational efficiency. Centralized analytics can automate root‑cause analysis and reduce manual reconciliation between inventory, network telemetry and OSS/BSS records.
- Personalized customer experiences. Cross‑domain analytics (usage, service quality, payment behaviour) enable more relevant offers and proactive retention interventions.
- Scalability and future AI workloads. A lakehouse foundation with Databricks speeds model training, deployment and inference at scale — an enabler for non‑interactive AI use cases (fraud detection, predictive maintenance) and interactive agentic AI in customer care.
- Workforce capability uplift. MTN’s heavy investment in Azure certifications — the company reports more than 1,250 employees certified under Microsoft’s Enterprise Skills Initiative — reduces operational risk during and after migration by building local skills.
The technical anatomy: what EVA 3.0 likely looks like (architectural sketch)
- Ingest layer: streaming collectors and batch connectors (Kafka/ Event Hubs / Data Factory) to capture network probes, call detail records, OSS/BSS events and app telemetry.
- Storage: Azure Data Lake Storage as the lakehouse store, with Delta Lake tables for ACID and time‑travel.
- Compute: Azure Databricks running Spark jobs and pipelines for ETL, streaming analytics and ML model training.
- Orchestration: Databricks Jobs/Workflows (or Azure Data Factory for certain flows) coordinating the ~800 analytic workflows reported.
- Security & governance: Microsoft Defender for Cloud, Entra ID for identity and role‑based access, and Purview (or equivalent) for data cataloguing and lineage.
- Serving & Ops: APIs and dashboards for downstream systems (OSS, CRM, NOC consoles) plus feature stores/serving endpoints for ML models.
Critical analysis — strengths, real value, and what to watch
Strengths and strategic upside
- Operational leverage: Consolidating telemetry into a governed, cloud‑native lakehouse unlocks automation that legacy OSS/BSS architectures struggle to deliver.
- Repeatability: If MTN truly templatizes EVA 3.0 for other operating companies, the group gains a powerful template for rapid analytics rollouts across markets.
- Skills and local capability: Heavy certification and the CCoE model reduce vendor‑operational friction and improve chances of long‑term success; MTN’s public reporting of 1,250+ Azure certifications is a meaningful indicator of investment in human capital.
- Vendor backing: Databricks and Microsoft are jointly deepening integrations on Azure; that improves platform maturity for large enterprise workloads.
Material risks and limitations
- Vendor lock‑in and architectural coupling. Building analytics around a single cloud + Databricks heavy stack accelerates development, but increases switching costs. Telcos should analyse multi‑cloud or hybrid fallback options and negotiate clear exit and portability clauses.
- Data sovereignty and compliance complexity. MTN operates across many African markets with differing regulations. Centralizing telemetry raises cross‑border transfer and retention questions; implementation must include region‑aware tenancy, encryption‑at‑rest and in‑transit, and robust audit trails.
- Security at scale. Defender and Azure tooling provide strong controls, but operationalizing security across millions of endpoints and billions of events is non‑trivial. Effective IAM, least privilege, CIEM posture (cloud entitlement management), continuous threat detection and runbooks for incidents are mandatory.
- Cost governance. Large‑scale Spark workloads and heavy storage can generate unexpected cloud spend. Proper tagging, quota control, cluster autoscaling policies and workload‑aware architecture are essential to avoid runaway costs.
- Observability and SRE maturity. Analytics platforms introduce their own operational failure modes. MTN must invest in SRE practices, error budgets, synthetic testing and automated remediation to keep the platform reliable under peak loads.
- Unverified scale claims. Key numeric claims (for example the widely reported “22 billion records per day” and the counts of workflows/data feeds) were published in coverage of the migration; however, those figures were not (at the time of writing) backed by a detailed third‑party benchmark or a public Microsoft case study that exposes throughput metrics, ingestion schemas or cost per terabyte. They should therefore be treated as MTN‑reported indicators of scale rather than independently validated performance metrics.
How to reduce risk: recommended operational and governance controls
- Create a multi‑tier data residency model:
- Deploy Azure regions or sovereign clouds aligned to local regulations for sensitive PII and call‑related telemetry.
- Use encryption keys stored in regionally‑scoped Key Vaults and implement granular key‑access policies.
- Harden Identity and Access:
- Enforce Microsoft Entra (Azure AD) conditional access, MFA and just‑in‑time privileged access for platform administrators.
- Adopt CIEM and continuous entitlement reviews to avoid privilege creep.
- Cost and resource governance:
- Implement workload quotas, per‑team budgets, and autoscaling policies for Databricks clusters.
- Use cost forecasting alerts and tag‑driven chargeback to incentivize efficiency.
- Security posture and incident playbooks:
- Ingest Defender alerts into a centralized SOC (or Sentinel) with runbooks that map detected issues to automated mitigations.
- Run regular red/blue exercises scoped to the analytics platform.
- Data quality, cataloguing and lineage:
- Implement data contracts, a formal schema registry and a catalog (Purview or similar) so downstream consumers can trust the data.
- Automate data quality checks in early pipeline stages and fail fast on schema drift.
- Responsible AI and model governance:
- Create a model registry with reproducibility requirements, bias testing, performance monitoring and a human‑in‑the‑loop gating process for production models.
- Hybrid/multi‑cloud exit planning:
- Standardize data export formats (open Parquet/Delta Lake) and document dependency graphs so workloads can be moved or recreated if required.
Practical checklist for other telcos planning a similar migration
- Assess telemetric volume and peak ingestion profiles; model costs for storage, compute and egress.
- Pilot a single high‑value use case (e.g., proactive NOC alerts) to validate latency and operational benefits.
- Build a CCoE and execute a certified training plan for engineers — certification numbers are not vanity metrics: skilled staff materially reduce migration risk. MTN’s 1,250+ Azure certifications (reported) exemplify this investment.
- Design for data locality and compliance from day one — not as a retrofitted control.
- Contract SLAs with hyperscaler and integrators that include throughput, availability and runbook obligations.
- Instrument end‑to‑end observability across streaming producers, ETL pipelines, model inference endpoints and serving APIs.
Business and regional implications for Africa’s digital ecosystem
MTN’s public positioning of EVA 3.0 as a blueprint for its African markets has broader implications beyond the company itself. If successful and repeatable, it accelerates a virtuous cycle of capability on the continent: telcos operating on a shared technical pattern can create more vibrant ecosystems for ISVs, BSS/OSS integrators, fintechs and government analytics projects. The move also places pressure on regional competitors to modernize or partner with large cloud vendors to stay competitive.At the same time, the public emphasis on certifications and workforce training highlights another imperative: cloud modernization must be matched by a local skills pipeline. MTN’s reported certification volume suggests an emphasis on building talent internally — a critical signal for governments and training bodies to scale cloud and security education across the region.
What remains to be seen (and should be requested/validated)
- Auditable performance benchmarks: independent throughput testing or a vendor‑published case study that details ingestion patterns, compression ratios, average query latencies and the cost per TB/day of sustained throughput.
- Governance artifacts: documented data retention policies per market, cross‑border transfer controls, and audit logs demonstrating how PII is protected in multi‑tenant analytics workflows.
- Operational metrics: incident MTTR improvements attributed to EVA 3.0, customer QoS improvements, and direct revenue or cost reductions tied to analytics use cases.
- Third‑party security assurance: results of independent penetration testing or compliance attestations that validate the Defender‑backed controls at scale.
Conclusion
MTN’s migration of EVA to Azure — coupled with Databricks and Defender‑class controls — is a textbook example of how large telcos are rearchitecting their operational core for the AI era. The technical choices reported (lakehouse + Spark + cloud security + a focus on workforce certification) align with industry best practice and the broader Microsoft telco strategy. Microsoft’s own telco messaging and Databricks’ continued platform investments make the architecture plausible and well supported. The concrete gains — faster insight, improved NOC responsiveness, and better personalization — are compelling. However, the most important next step for MTN, its partners and customers is transparent, auditable validation: independent performance benchmarks, clear data‑sovereignty design, and rigorous security certification. Until those are visible in published technical case studies, the large‑scale throughput statistics should be regarded as MTN‑reported figures that require corroboration for engineering or procurement decisions.For African telcos, MTN’s EVA 3.0 represents both an operational template and a cautionary tale: cloud platforms can unlock dramatic new capabilities, but they must be paired with robust governance, security, cost controls and people capability to deliver sustainable value.
Source: innovation-village.com MTN, Microsoft Launches Africa’s Largest Telco Cloud Platform - Innovation Village | Technology, Product Reviews, Business
