Telcos have massive amounts of data. So why are we failing at AI? In a word: context. Historically, telecom systems treated network elements like cell sites as isolated technical objects. AI can’t work like that. When you redefine network elements as business entities connected to subscribers, traffic, infrastructure, and revenue, you can correlate information across OSS, BSS, and network systems in real time. That semantic consistency is exactly what AI needs to transform operations. That’s how Zain Sudan reduced diagnostic time for dead cell sites from 48 hours to 30 minutes. Want that kind of improvement? First, give yourself the foundation with an ontology. Then, watch your AI take off!
NEC closed its $2.9B acquisition of CSG, folding it into Netcracker. Look at the math: CSG’s $1.24 billion trailing revenue = a 2.3x multiple. Modern SaaS trades at 6–10x or more. Why is another BSS vendor pricing like a services business? Oh yeah, because it is one. The “software” line lags behind the consulting, customization, and managed services that actually generate the revenue. And AI is about to gut that consulting layer. When agents do in hours what consultants do in months, the services revenue evaporates—and 2.3x starts looking generous. Looks like Netcracker overpaid. 💸
Verizon joined Anthropic’s Project Glasswing as the first telco with access to Claude Mythos Preview—the model that found a previously undetected 27-year-old OpenBSD vulnerability. Smart move, but what code will Verizon evaluate? Telcos mostly run vendor solutions to which they have zero source-code access. Verizon can scan whatever it built in-house, but the Ericsson, Nokia, Cisco, and Samsung stacks running the RAN, core, transport, and OSS are black boxes. If a 27-year-old zero-day bug has been sitting around in open code, what’s lurking in the closed code running your network? You can’t audit what you can’t read. Will Verizon pull in Ericsson for the code review of its life?
AWS Interconnect – last mile just hit general availability (GA), letting enterprises spin up private, high-speed AWS connections from any branch or data center in a few clicks. Lumen is the launch partner; AT&T followed at MWC with fiber and fixed wireless. This is the right hyperscaler bet for telcos—and a smarter one than chasing sovereign AI clouds and becoming GPU landlords. Enterprises moving AI to production need secure, private, low-latency pipes into hyperscaler environments. That’s a problem telcos can actually solve: fiber, SLAs, encryption, and an enterprise relationship hyperscalers don’t have. You can’t beat AWS on compute economics. But you can sell them the pipes that make AI workloads enterprise-grade. Lean into connectivity for AI. Skip the GPU fantasy.
Contrast that with this: Deutsche Telekom, Orange, TIM, Telefónica, and Vodafone are recruiting more telcos to join their “European Edge Continuum,” a federated sovereign edge platform across Europe. The current five only cover ~55% of the European population, so they want more operators to join. The pitch: a unified pan-European SLA so enterprises don’t have to negotiate individual contracts with multiple operators across different countries. This is exactly the architecture AWS just collapsed to a few console clicks via Lumen and AT&T. The five operators are still recruiting members and validating use cases in a lab; AWS shipped GA. The sovereign-edge consortium model assumes enterprises will accept a worse, slower-to-build version of what hyperscalers already deliver, in exchange for an EU flag on the data center. They won’t. Enterprises will pick the productized option every time. Build the pipes into hyperscaler regions. Don’t try to BE the hyperscaler.
AT&T, T-Mobile, and Verizon just announced a joint venture to pool spectrum on a unified direct-to-device satellite platform, framed as ending rural dead zones. Carriers that compete on everything else are united by one threat: Musk. But Musk isn’t trying to beat telcos; he’s monetizing Starlink to fund his mission to Mars. And he’ll find a telco that helps him, so watch for which non-US number two or three blinks first: one with brutal rural geography, coverage pressure, and enough subscribers to make it worth the while for SpaceX. That carrier gets coverage; Musk gets the rocket budget. The joint venture doesn’t stop that.
USA Today profiled Goji Mobile (a TelcoDR company) this week—a consumer marketplace comparing 100+ phone plans with one-click switching. AI is making consumers smarter, and we’re already seeing it in the actions users take on Goji: 74% pick an MVNO over Verizon, AT&T, or T-Mobile. The Tier-1s saw this coming, which is why T-Mobile owns Mint, AT&T owns Cricket, and Verizon owns Visible, but owning the alternatives doesn’t fix the pricing problem. Bundle confusion, contracts, and store visits don’t survive a 2-minute, side-by-side comparison plus eSIM switching. Consumers are already asking ChatGPT and Claude which plan is best–and that’s a 10-second answer. You can’t ride postpaid coattails forever. Goji is showing telcos what you can do when you start catering to subscribers individually. Give it a try! 🍊
