How Telcos Become Tech Firms: LMT’s AI Pivot

Communications now represents 50% of LMT Group’s business volume—the other half comes from areas like IoT, defence, retail & logistics, finance, and IT solutions. That’s not a “nice adjacent revenue stream.” That’s a company reshaping its identity.

For telco leaders reading this (especially in smaller markets), LMT’s story lands at the right moment. It’s December 2025, budgets are being finalised, and the same question keeps showing up in boardrooms: how do we grow when connectivity is already a mature business? LMT’s answer is blunt: stop behaving like a connectivity-only operator and start acting like a technology company—using AI in telecommunications as a practical engine for scale.

This post breaks down what LMT changed, why the “telco-to-tech” pivot is happening now, and how AI and 5G management actually fit into a strategy that can generate new products (not just new integrations).

LMT’s pivot in one line: products beat pipes

LMT’s strategic move is clear: treat the network as a platform, then build exportable products and solutions on top of it. The organisation has grouped those bets into a multi-sector ecosystem—LMT IoT, LMT Defence, LMT Systems, LMT Retail & Logistics, LMT Finance, LMT Innovations, and Santa Monica Networks.

Here’s the stance I agree with: being “not just a pipe” isn’t a slogan. It’s a portfolio decision. If you’re still measuring your future by subscriber growth and ARPU alone, you’re choosing to compete in the most price-sensitive layer of the stack.

LMT’s leadership frames the constraint honestly: a relatively small European market limits the customer base. So growth has to come from diversification and exportability—solutions that travel beyond Latvia.

What changed operationally (not just in a press release)

A pivot like this isn’t real until three things happen:

• Revenue mix shifts (LMT’s is now 50/50 between communications and new directions).
• Capabilities shift from integration-heavy IT services toward physical products.
• R&D and go-to-market becomes multi-vertical (defence, smart cities, IoT, mobility) instead of telecom-only.

That last point matters because it forces a different kind of engineering and product discipline—especially if you want AI-driven services that work reliably at scale.

Why 5G created pressure… and AI turns it into margin

LMT points to the introduction of 5G as a “seminal moment.” The subtext is familiar across the industry: 5G shipped with big industrial promises, but most customers didn’t see immediate outcomes. That gap (promise vs. proof) is where many telcos got stuck.

LMT’s response was to turn that gap into an innovation trigger: when the market needs something real and it doesn’t exist yet, build it.

Here’s where AI in telecommunications stops being theoretical and becomes a business tool.

AI isn’t the product—AI is the multiplier

Most operators pitch AI as automation. That’s fine, but it’s incomplete. AI is also what makes a telco’s new products:

• cheaper to run (less manual operations)
• faster to adapt (model-driven optimisation)
• more reliable (predictive maintenance and anomaly detection)
• more scalable (repeatable decision systems)

If you’re building exportable offerings—IoT platforms, managed services, defence-grade situational awareness—your margins depend on how much you can standardise and automate. AI is how you standardise decisions.

Where AI shows up in a telco-to-tech pivot

In practical terms, the AI-heavy layers that matter most are:

1. Network optimisation: traffic prediction, cell parameter tuning, congestion mitigation.
2. 5G management: slice assurance, QoS prediction, RAN anomaly detection.
3. Edge intelligence: on-device or edge AI inference to cut latency and backhaul.
4. Service assurance: incident correlation and root-cause analysis across domains.

The “tech company” identity becomes believable when these AI capabilities are packaged into products customers can buy—and when the operator can run them without throwing headcount at every deployment.

The IoT Shortcut: a small device with big strategic intent

LMT Group’s newest product is the IoT Shortcut, positioned as the smallest mobile data module in the world for sensor equipment, designed to give manufacturers universal connectivity options.

The headline feature is tiny hardware. The more strategic feature is what sits around it:

• software that automates data processing and integration
• connectivity automation
• cloud data services
• a customisable customer platform and application

That’s the blueprint for “telco as product company”: the module is an entry point, but the recurring value sits in the software and operations layer.

Battery life is a business feature, not an engineering flex

LMT claims very low energy consumption, enabling sensor battery life up to 10 years. In IoT buying decisions, long battery life does three things:

• lowers maintenance visits (often the biggest lifetime cost)
• increases deployment viability in remote/industrial sites
• improves customer ROI calculations enough to get projects approved

If you’re selling IoT into municipalities, construction, logistics, or environmental monitoring, the procurement conversation is rarely about throughput. It’s about operational cost and reliability.

Proof points: early partners and real-world use cases

The module is being tested with partners including:

• a wireless sensor manufacturer
• bark beetle activity monitoring software (forestry/environment)
• a construction monitoring specialist

This diversity matters. It suggests the product is being shaped for multiple verticals, not just a single niche.

Edge AI + IoT: why the Infineon angle matters

LMT’s team is negotiating planned cooperation with a major European chip manufacturer to integrate the module into Edge AI solutions.

This is the most forward-looking part of the story because it points to an architecture telcos should be betting on:

Do more inference at the edge or on the device, transmit less data, and reserve the network for what actually needs transport.

LMT highlights three expected outcomes from local, on-device AI processing:

• reduced data transmission volume
• faster reaction speed
• extended battery life

That trio maps directly to competitive advantage in large-scale sensor deployments.

Why edge AI is becoming the default for IoT economics

If you’re managing thousands (or millions) of devices, sending every raw sample to the cloud is an expensive habit:

• it burns radio resources
• it increases cloud compute/storage costs
• it introduces latency
• it creates privacy and data residency issues

Edge AI shifts the model from “ship all data” to “ship only events, anomalies, and summaries.” That’s not just good engineering. It’s how you keep IoT unit economics from collapsing at scale.

What other telcos can copy from LMT (and what they shouldn’t)

LMT is a useful case study because it reflects constraints many operators share: limited domestic market size, intense competition, and pressure to justify 5G investments.

But copying the shape of the pivot isn’t enough. The details matter.

Copy this: build around unmet needs, not internal capabilities

LMT’s VP describes repeatedly encountering critical unmet market needs that “force us to build the solutions ourselves.” That’s the right direction.

Here’s a practical filter I’ve found works when evaluating diversification ideas:

• If the “product” requires constant custom integration, it’s probably a services business.
• If you can package it, price it, support it, and deploy it repeatedly, you’re closer to tech-company economics.

Copy this: treat AI as an operations layer and a product feature

If AI only lives inside your NOC, it caps out as cost-saving.

If AI also lives in your products—edge analytics, anomaly detection, predictive maintenance, automated provisioning—it becomes revenue.

A simple way to plan this is to map AI in two columns:

• Run the network better (network optimisation, 5G management)
• Sell outcomes (IoT assurance, edge AI insights, security monitoring)

The same models and data pipelines often serve both.

Don’t copy this blindly: “multi-vertical” without focus

A multi-sector group structure can accelerate learning, but it can also create a messy portfolio. The risk is launching too many initiatives without a shared technical backbone.

If you want the diversification to compound, you need reuse across:

• identity and device management
• observability and incident management
• data governance
• MLOps (model monitoring, retraining, drift detection)

Without that, every vertical becomes a separate bespoke system, and the organisation slowly turns into an integration shop.

A practical blueprint: “telco to tech” with AI in telecommunications

If you’re leading strategy, product, or technology in a telco, here’s a concrete way to apply LMT’s pattern over the next two quarters.

1) Choose one exportable problem statement

Examples that fit telco strengths:

• uptime assurance for industrial IoT fleets
• edge video analytics for smart city safety
• private 5G + device management for logistics hubs

One is enough. The goal is repeatability.

2) Build a reference architecture that assumes AI from day one

Not “we’ll add AI later.” Start with:

• telemetry standards
• data retention rules
• model monitoring
• human-in-the-loop escalation paths

AI that can’t be operated safely becomes shelfware.

3) Tie network optimisation and product KPIs together

This is where operators have an unfair advantage.

If your product depends on latency, availability, or battery life, your 5G management and network optimisation teams can directly improve product outcomes—then measure it.

A strong KPI trio for IoT-like products:

• device attach success rate
• time-to-detect anomaly (edge or cloud)
• field maintenance visits avoided

4) Package support like a tech company

If customers can’t understand onboarding, pricing, and support boundaries in one page, you don’t have a product yet.

Write the runbook before you scale the sales.

Where this is headed in 2026

The telco industry’s most reliable growth path right now is AI-enabled specialisation: pick domains where you can combine network assets, engineering, and operational discipline into outcomes customers will pay for.

LMT’s pivot is a strong example because it treats 5G as a platform and AI as the operating system for scale—then backs that up with product moves like IoT Shortcut and edge AI partnerships.

If you’re planning your 2026 roadmap, the question isn’t “Should we do AI in telecommunications?” You’re already doing it somewhere. The real question is: are you using AI only to run networks cheaper, or also to build products that can grow beyond your footprint?

If you want a second set of eyes on where AI can create revenue (not just savings) in your telecom portfolio—network optimisation, 5G management, IoT assurance, or edge AI—let’s talk. The fastest wins usually come from tightening one workflow end-to-end, then packaging it as something customers can actually buy.