AI Telecom Winners 2025: What the Awards Reveal

350 billion data points a day. That’s the scale one award-winning telco AI platform is already processing in production—and it’s not a lab demo or a slide deck.

The 2025 World Communication Awards (WCAs) matter for a simple reason: they’re a real-time snapshot of what’s working in telecom right now. Not what’s trendy, not what’s “coming soon”—but what’s delivering measurable outcomes across networks, security, cloud, customer experience, and new services.

For this AI in Telecommunications series, I’m using this year’s winners as a filtering mechanism. When an operator wins for fewer outages, lower OPEX, higher ARPU, or threat prevention at scale, that’s a signal. And the signal is loud: AI is moving from “nice to have” to “operating system.”

The 2025 award list points to one big shift: AI is now operational

AI in telecom isn’t a side project anymore; it’s being judged (and rewarded) as a core capability that improves reliability, costs, security, and customer outcomes.

Look at the strongest submissions in 2025 and you’ll notice a pattern: winners didn’t win by “using AI.” They won by packaging AI into operating processes—then proving results.

A few examples from the WCAs that show how mature the market has become:

• AI Innovation winner: Jio Platforms (JioBrain) processing 350B data points daily, reporting 40% fewer outages, 30% lower OPEX, and 20% higher ARPU across 200+ million 5G users.
• 5G Award winner: Singtel 5G+, turning network slicing into everyday customer value and integrating Security-as-a-Slice, actively blocking 6.6 million threats per month.
• Total Experience winner: Sparkle, combining AI-based automation and real-time tools with a deliberately human service model.

If you run network operations, IT, digital, or product in a telco, this matters because it changes the benchmark. The bar is no longer “we piloted AI.” The bar is “we shipped AI into the workflow and can defend the numbers.”

What award-winning AI in telecom is actually doing (not promising)

The WCAs cover many categories, but the telecom AI story clusters into a few practical jobs. These are the areas where AI is creating repeatable advantage.

1) Network assurance that prevents incidents—not just detects them

The goal in modern network operations is fewer incidents per subscriber, not faster war rooms.

JioBrain’s reported 40% outage reduction is the kind of outcome you only get when AI is connected to:

• near-real-time telemetry (radio, core, transport, devices)
• automated root-cause analysis
• change management and closed-loop actions
• consistent operational governance (so humans don’t override the system every time)

If you’re building an AI network operations strategy in 2026, prioritize two deliverables:

1. Prediction you can operationalize (alerts tied to actions, not dashboards)
2. A feedback loop (did the action help, hurt, or do nothing?)

Without those, you’re collecting “interesting insights” while outages keep happening.

2) AI-driven cost control that doesn’t sabotage customer experience

Operators keep getting asked to cut costs while supporting more traffic, more devices, and more service complexity. That’s exactly where AI belongs—because humans can’t manually tune systems at that scale.

The “30% lower OPEX” outcome reported by the AI Innovation winner should be read carefully: it’s not just labor reduction. It usually comes from a mix of:

• fewer truck rolls via predictive maintenance
• fewer repeat incidents via systemic fixes
• better energy optimization across RAN and data centers
• reduced time-to-resolve through automated triage

Here’s the stance I take: if your AI cost program harms NPS, it’s not optimization—it’s debt. The winners show that cost and experience can improve together when automation is paired with clear guardrails.

3) Security that’s integrated into the product, not bolted onto the network

Singtel’s Security-as-a-Slice and monthly 6.6M threats blocked is an important clue about where telco security is heading: toward customer-facing security offers built into connectivity.

That trend is reinforced by the WCA focus on quantum-safe security (multiple winners and finalists across categories). Quantum-safe initiatives are partly about cryptography—but commercially they’re also about trust and differentiation.

If you’re designing AI-powered telecom security services, the winning pattern looks like this:

• Security packaged as an add-on or tier (easy for customers to buy)
• Security delivered with network controls (not just endpoint apps)
• Security outcomes measured (threats blocked, time to mitigate, false positives)

Customers don’t want more alerts. They want fewer bad days.

5G is “coming of age,” and AI is the reason it can scale

The WCAs 5G category commentary was blunt: 5G is finally maturing. That doesn’t happen just because radios improved. It happens because operating 5G profitably requires software intelligence—especially as slicing, private networks, and MEC expand.

Network slicing is becoming a product (and AI helps keep the promise)

The interesting part of Singtel’s 5G+ win isn’t that slicing exists; it’s that the judges called out “democratization of consumer network slicing.” That means telcos are getting closer to selling slices like customers understand them: priority, security, performance guarantees.

AI is what keeps that from turning into an operational mess:

• forecasting demand per slice
• anomaly detection per slice (not just the whole network)
• policy optimization to prevent slice contention
• automated assurance reporting (so enterprise customers trust the SLA)

If you’re in product or B2B, this is a practical takeaway: you can’t monetize slicing if you can’t assure it. Assurance requires AI because the variables outnumber human attention.

Private 5G + AI is becoming the default enterprise pitch

The Enterprise Service of the Year winner—a smart wind farm private 5G network with AI-driven interference management—captures a wider market reality: enterprise wants connectivity plus outcomes.

That’s why private 5G keeps clustering with AI:

• predictive maintenance (equipment + network)
• safety monitoring and incident prevention
• quality optimization in harsh RF environments
• site-level autonomy where backhaul is constrained

Enterprises aren’t asking for “a SIM card.” They’re asking for operational reliability.

Cloud, digital transformation, and CX: the “plumbing” that makes AI pay off

AI projects fail for boring reasons: fragmented systems, inconsistent data, and teams that can’t ship.

That’s why the WCA categories around cloud and digital transformation are more connected to AI than they look.

Digital transformation winners are quietly solving the AI prerequisite: clean execution

The Best Digital Transformation Programme winner migrated 88 million subscribers to a unified digital monetisation platform. That’s not just IT hygiene. It’s what enables:

• consistent customer journeys (fewer broken handoffs)
• unified product catalogs (faster experimentation)
• standardized event streams (better ML training data)
• reliable billing and entitlements (fewer CX failures)

In practice, your AI CX automation is only as good as your BSS/OSS foundations. If those foundations are fractured, AI becomes a bandage.

Cloud platforms and private MEC are where telecom AI gets deployed at scale

The Cloud Award winner highlighted cloud platforms and private MEC. That’s relevant because many high-value AI use cases (video analytics, industrial monitoring, low-latency assurance) want computation closer to where data is produced.

The telecom-specific reality is this: edge AI isn’t about being fancy—it’s about meeting latency, sovereignty, and cost constraints.

If you’re planning MEC + AI for 2026, focus on two questions:

1. What inference needs to be local? (latency, privacy, bandwidth)
2. What training belongs centrally? (cost efficiency, broader datasets)

Getting that split right is often the difference between a profitable deployment and an expensive science project.

CX winners show a balanced model: automate the routine, protect the human moments

Sparkle’s Total Experience win is a good corrective to a common mistake: over-automating customer experience.

AI should handle:

• intent detection and routing
• proactive notifications (outage, billing, usage anomalies)
• summarization for agents
• next-best-action recommendations

Humans should handle:

• exceptions and escalations
• retention conversations
• high-risk security and fraud cases
• enterprise incident communications

A simple metric I’ve found useful: if your automation lowers average handle time but increases repeat contact, you didn’t improve CX—you shifted the pain.

A practical checklist: how to copy the winners without copying their budgets

Not every operator has Jio-scale data or Singtel-scale product teams. You can still adopt the same operating principles.

Start with one measurable outcome and design backwards

Award-winning telecom AI is built around a number you can defend. Pick one:

• outage minutes reduced
• first-time-right change rate
• mean time to detect / resolve
• energy per GB
• fraud loss rate
• call deflection with stable CSAT

Then design data, models, and workflows around that outcome.

Build an “AI to operations” path (the missing middle)

Many teams stop at model accuracy. Winners go further: they operationalize.

Your path should include:

• alert quality thresholds (precision/recall with business impact)
• playbooks for actions
• human approval gates for risky actions
• audit logs and rollback
• post-incident learning (model + process updates)

Treat telecom AI governance as a product feature

In 2026, buyers will ask how your AI behaves under stress and scrutiny.

Governance that actually helps includes:

• model monitoring for drift
• data lineage and access controls
• explainability for regulated decisions (fraud, credit, security)
• clear accountability when automation acts

This is especially true as quantum-safe security and critical infrastructure requirements gain momentum.

Where this is heading in 2026: AI becomes the telco operating system

The WCAs 2025 winners show the direction of travel: AI is becoming embedded across the stack—RAN, core, cloud, BSS, security, and customer operations.

The operators that win the next wave won’t be the ones with the most pilots. They’ll be the ones that turn AI into repeatable operating capability—and can prove it in numbers customers and CFOs care about.

If you’re mapping your AI in Telecommunications roadmap for 2026, here’s the next step I’d take: pick one high-impact domain (assurance, security, or CX automation), define the measurable outcome, and build the workflow so the model can actually act.

What would change in your business if you could credibly promise 40% fewer outages—or if your security offer could show customers the threats you blocked last month?