Nokia just announced its third major pivot in three years—from consumer phones to B2B enterprise (MWC23), and new CEO Justin Hotard is compressing four business units into two while exiting long-time mobile chief Tommi Uitto. The new tagline shifts from “connecting people” to “connecting intelligence,” banking on network infrastructure as the “AI supercycle” growth engine to hit €2.7-3.2 billion operating profit by 2028. The stock dropped 8.8% because investors remember Hotard’s last AI pivot—running Intel’s data center unit into the ground while NVIDIA ate their lunch. Maybe Justin thinks Data Center Part Deux will work better this time around? For telco execs watching your primary RAN vendor chase hyperscaler dreams: when your equipment supplier keeps reinventing itself every 18 months, you’re the one left holding the integration risk while they figure out what business they’re actually in.
Benedict Evans just dropped his latest 90-slide The State of AI. Evans’ key insight? Unlike mainframes, PCs, the web, or smartphones, we genuinely don't know how much better LLMs will get—which is why Microsoft, AWS, Google, and Meta are spending $400 billion on AI infrastructure in 2025 (more than the entire global telecom industry's $300 billion annual capex). Google CEO Sundar Pichai says “the risk of under-investing is dramatically greater than over-investing” and Meta CEO Mark Zuckerberg shrugs that the “very worst case” is they’ve “just pre-built for a couple of years.” Translation for telcos: the hyperscalers are placing bets with house money and acceptable downside, while operators still run 18-month RFPs for every purchase. Stop waiting for AI use cases to “become clear”—start building now, or get left behind.
Google dropped Gemini 3 on November 18th with PhD-level reasoning benchmarks and claims of being the “most intelligent model yet” crown firmly on its head. Six days later, Anthropic released Claude Opus 4.5 and snatched it right back with 80.9% on SWE-bench Verified (first model ever to break 80%!), reclaiming the coding throne. This follows OpenAI’s GPT-5.1 on November 12th and xAI’s Grok 4.1 the same week—four frontier models in two weeks. For telco executives still debating which AI vendor to “partner” with: you’re asking the wrong question. When the best model changes weekly and pricing has collapsed 70% in a year (Opus went from $15/$75 to $5/$25 per million tokens), the models themselves are commoditizing fast. Build your AI stack to swap models like you swap suppliers—because loyalty to any single LLM is a strategic liability, not an advantage.
Two telcos just made the right call on the public cloud: Charter partnered with AWS for AI-powered developer tools across Spectrum services. Vodacom inked a multi-year deal with Google Cloud for data analytics, network optimization, and fraud detection. Both recognized that hyperscalers have already spent billions on infrastructure, security, and AI—so why replicate it when you can rent it? The strategic edge for telcos isn’t in running your own cloud; it’s in applying hyperscaler capabilities to the customer data and network assets only you possess. Let AWS and Google handle the undifferentiated heavy lifting while you focus on winning subscribers.
T-Mobile unveiled its 15-minute switching experience via the T-Life app—AI-powered “Easy Switch” recommends the best plan based on your current AT&T or Verizon account, and gives you 90 days to decide on a new phone. Verizon’s whiny response? Speed “prioritizes the transaction over the quality of the experience” and AI “might not be best suited to make choices for people.” Translation: T-Mobile is building a customer experience that removes every reason not to switch, while Verizon is praying customers are too lazy to leave. With 65% of Americans saying switching is a hassle, one carrier sees that as an opportunity to get new subs and the other sees it as a moat to trap customers. That tells you everything you need to know about who’s playing offense and who’s playing defense.
New Zealand’s (and Totogi customer!) 2degrees just launched a free 7-day eSIM trial—15GB data, 500 minutes, 100 SMS, no credit card required. CEO Mark Callander says they’re “so confident in our network that we want people to experience it for themselves”. Gotta love the swagger! Here’s the insight: coverage maps are marketing documents, not reality. They can’t tell you if you’ll have a strong signal at your home, your commute, or your kid’s school. The only way to know if a network works for you is to actually use it. At Goji (last week’s podcast!) we built hyperlocal coverage maps for all three US networks that show real performance at real addresses. People love it. Maybe Verizon should look into this approach, too…
Deutsche Telekom and Orange CEOs issued a joint statement forming the European Sovereign Tech Industry Alliance (ESTIA) with 10 partners, calling for a “Digital Networks Act” to modernize regulations that supposedly hamper their €50 billion annual network investment. They want the EU to mandate European cloud services in public procurement because, you know, AWS, Azure, and Google Cloud control over 70% of the European market and that share keeps growing. Here’s my question for these guys: remind me what happened to Gaia-X? That 2019 initiative was supposed to deliver European digital sovereignty too, but got mired in bureaucracy, infighting, and—ironically—ended up including the very hyperscalers it aimed to challenge. European telcos love blaming regulations while US hyperscalers eat their lunch, but maybe the problem isn’t the rules— it’s the committees and manifestos that don’t ship products.
I missed promoting Microsoft Ignite when it happened November 18-21 in San Francisco—my bad (sorry Rick!). But the good news: most of the 400+ sessions are available on-demand, including some solid telco content. Check out “Unleashing Telco Innovation: Agentic AI & Connected Intelligence,” “Securing Private Wireless: From Design to Deployment” and one I’d particularly recommend: “Accelerate IT transformation with agentic AI,” where AT&T discusses partnering with Microsoft and IBM to deploy multi-agent AI systems across their IT operations. Worth a watch if you’re mapping out your 2026 AI roadmap on the Microsoft platform!
Verizon announced plans to cut around 13,000 workers, its largest layoff ever, while converting nearly 180 stores to franchises under new CEO Dan Schulman’s restructuring. Brutal but necessary—when you’re hemorrhaging customers to T-Mobile and cable operators, something’s gotta change, and change big. Schulman said the cuts aim to “reorient” the entire company toward delighting customers. The key area he’s cutting? ENTERPRISE NETWORKS. Which reminds me, thank goodness I didn’t buy CASA Systems! Wonder how that’s working out for Lumine Group? Probably not as good as they hoped with that $32.3M price tag…😅
