All I want for Christmas are these 5 things from telcos
Confession: my favorite Christmas song is Mariah Carey’s “All I Want for Christmas Is You.” Singing it on repeat got me thinking about what I really do want this year. So I made this list:
Consider it my holiday gift to you: practical advice that could save you millions of dollars and years of wasted effort. I didn’t just make this up. My list is based on what I’ve seen work (and not work) with real telcos trying to deploy AI and the public cloud across their organizations.
1. Give up the RFP process to choose AI vendors.
Requests for proposals made sense when software capabilities were static and innovation cycles were measured in years. But AI is different. This technology evolves weekly, not annually. By the time you’ve slogged through 100-page responses from a dozen vendors, the capabilities you were evaluating will be two generations behind.
Think about what’s happened just this year: Claude went from Sonnet to Opus, OpenAI shipped o1 reasoning, and Google released Gemini 2.0—each one changing what “good” looks like. Your RFP criteria from Q1 are already measuring the wrong things.
RFPs force you to specify requirements before you understand what’s possible. With AI, the best solutions often come from capabilities you didn’t know to ask for. You can’t RFP your way to innovation.
Here’s a better approach: Run 30-day proofs of concept with your actual data. Not demos, not PowerPoints, but real implementations solving real problems. Give three vendors access to your data and a specific challenge. If they can’t show measurable value in 30 days, they won’t deliver it in three years either.
At Totogi, we built an entire BSS in one day at The AI-Native Telco Forum in Düsseldorf. One day. That’s the speed AI enables when you stop talking and start building. The best AI vendors aren’t afraid of short trials. In fact, they can’t wait to show you what they can do. The ones pushing for long RFP processes are probably hiding something.
Read more: Stop procuring AI. Start using it.
2. Build agents with an underlying ontology.
Everyone’s racing to deploy AI agents. Most will fail. Not because the AI isn’t smart enough—but because nobody taught it what your business data actually means.
Your telco runs on hundreds of systems, each with its own definition of “subscriber,” “plan,” and “product.” Your IT team burns endless hours on integrations and APIs, trying to force these systems to speak the same language. That’s semantic inconsistency—and it’s been eating your budget for decades.
An ontology fixes it at the root. One authoritative source that every system—and every AI agent—queries for the truth.
Here’s why this matters for AI: without an ontology, you’re asking agents to learn your business logic in the agentic layer. You end up with 100 agents learning 20 different definitions. When rules change, someone has to update every agent. With an ontology, you impose business logic once and every agent inherits the change automatically.
Building a proper ontology takes time and expertise. At Totogi, where I’m acting CEO, my team and I built one in BSS Magic. It was worth the effort because it’s the foundation that makes everything else work.
Read more: Why your agentic AI strategy will fail
3. Skip the data lake.
Your data lake project is solving the wrong problem.
You’re running your business on your existing systems right now. Today. Billing is billing. CRM is CRM-ing. Network is networking. The data is clean enough to run a multi-billion dollar operation. So why are you spending three years and tens of millions trying to copy it all into one place?
Because someone told you the problem was dirty data. It’s not. The problem is semantic inconsistency—your systems don’t agree on what “subscriber,” “product,” or “churn” actually mean. And a data lake doesn’t fix that. It just moves the inconsistency somewhere more expensive.
An ontology solves the actual problem, because it’s one authoritative definition layer that sits on top of your existing systems. No migration. No three-year project. Just semantic consistency—the thing you actually needed all along.
Stop waiting for the lake. Start building AI with the data you already have.
Read more: Context is EVERYTHING
4. Stop using “sovereignty” as an excuse to avoid the cloud.
I hear it constantly: “We need to keep data on-prem for sovereignty reasons.” But here’s what that really means: you’re about to spend two years building infrastructure instead of building AI.
Two years procuring H200s. Two years hiring a team you can’t afford to retain. Two years standing up infrastructure that’ll be obsolete before it’s operational. Meanwhile, your competitors picked a cloud region and started shipping AI products on day one.
And here’s the kicker: those H200s you’re procuring? Huang’s Law says AI chip performance improves 8–15x every two years. Your shiny new infrastructure will need a full refresh before you’ve even finished deploying it. On a public cloud, you get the latest chips the moment they ship. On-prem, you get a perpetual procurement cycle.
Because here’s the thing: every major hyperscaler already offers sovereign cloud regions. AWS, Google, and Azure all have regions designed to meet the strictest data residency and regulatory requirements—Europe, Asia, the Middle East, wherever your regulators demand. The compliance box is already checked. You just have to pick a region.
Sovereignty isn’t a reason to build your own AI datacenter. It’s an excuse to delay. Pick a region and start building.
Read more: Move over, Moore. It’s Huang time.
5. Don’t protect what’s already broken.
Your biggest competitor isn’t other telcos or hyperscalers. It’s the procurement process that takes 18 months, the security review that takes six, the architecture board that meets quarterly, and the vendor who swears migration will take three years.
I’m going to say the quiet part out loud: your systems are already broken. You’ve just gotten so good at working around them that you’ve mistaken workarounds for functionality. Your customer experience is already fractured; you’ve just normalized high churn as “industry standard.” Your data is already inconsistent; you’ve just hired armies of analysts to manually reconcile it.
So what exactly are you protecting?
The reason nobody touches that system isn’t technical. It’s political. If it breaks on your watch, you own it. But here’s the thing: it’s already broken. You’re just not getting blamed for it yet.
Pick one legacy system everyone’s afraid to touch—and kill it. Not a study. Not a pilot. Kill it. Replace it with something built in 30 days. You’ll learn more from that one act than from two years of transformation planning.
Do you think that’s too risky? We built an entire BSS in one day at Düsseldorf. Your legacy system isn’t more complex than that. AI has inverted the risk calculus. Today, the biggest danger isn’t trying something new and failing. It’s not trying anything new at all.
Read more: Düsseldorf changed everything for BSS
Adding it all up
The five things on this list aren’t revolutionary. They’re practical steps that could transform your business in 2026. But there’s no Santa Claus for this stuff. Fulfilling these wishes takes real-world action.
Notice the pattern? Every item on this list is about speed. Speed to evaluate. Speed to build. Speed to market. The telcos that win in 2026 won’t be the ones with the best strategy decks. They’ll be the ones that moved first.
It’s an important moment in the telco industry. AI is real, cloud is proven, and the opportunity is massive—but only for those willing to grab it. I’m going to make one more Christmas wish: that one telco exec reads this, picks the easiest item on the list, and starts before the champagne goes flat on New Year’s Eve.
Because the best time to start was three years ago. The second-best time is before your competitors finish reading this same post.
Happy Holidays. Now go break something! 🥚
Recent Posts
Get my FREE insider newsletter, delivered every two weeks, with curated content to help telco execs across the globe move to the public cloud.
Get started
Contact Totogi today to start your cloud and AI journey and achieve up to 80% lower TCO and 20% higher ARPU.
Learn
Telcos have one of the most semantically fragmented environments in any industry. This blog post explains how a telco ontology decodes the chaos and unlocks the value of AI.
Engage
Set up a meeting to learn how the Totogi platform can unify and supercharge your legacy systems, kickstarting your AI-first transition.
Believe
See how Totogi built a fully working, production-ready BSS—live, on stage—in just one day. We proved that our product, BSS Magic, really works like we say it does.