Telco Magazine just published its top 10 most disruptive telco startups for 2026. AST SpaceMobile took #1 for direct-to-device satellite. Totogi landed at #4. I’ll take it. AST is connecting people who have literally zero coverage—hard to argue with that. But it’s worth noting what the rest of the list looks like: the startups reshaping telco aren’t building better versions of legacy products. They’re building around them, with AI-first principles at the core. That’s the structural shift. The next few years are going to be fun. 🙌
Iranian missiles hit the headquarters of Batelco, Bahrain’s largest telecom operator, which also hosts Amazon Web Services (AWS) infrastructure. This attack follows the earlier drone strikes on AWS data centers in the UAE. Data centers are now military targets. That’s not a theoretical risk scenario; it’s happening right now, in real time, to real operators. The operators who added a public cloud disaster recovery (DR) layer in a different region are failing over. The ones who didn’t are calling their vendors, who are quoting six-to-nine-month recovery timelines. This is exactly why I’ve been saying your DR needs a DR. The public cloud lets you move your most critical systems out of the affected region entirely—in minutes, not months. If you’re a telco in the Middle East and you haven’t set up public cloud DR yet, what are you waiting for? Totogi can help. Give us a call!
SoftBank’s Large Telecom Model (LTM) just achieved a top‑tier overall ranking among 84 models on the GSMA’s new Open-Telco LLM Benchmarks, which test spec comprehension, log interpretation, and domain Q&A. It’s a credible technical achievement. But Wharton’s Ethan Mollick has been warning that domain-specific models are on the wrong side of what AI researchers call “the bitter lesson:” general-purpose models keep absorbing vertical knowledge faster than specialist models can stay ahead. The benchmarks SoftBank topped today will look different in 12 months when frontier models have ingested the same telecom training data with 100x the compute budget. The real moat in telco AI isn’t a better model. It’s an operational context layer, an ontology, that defines what your business means and constrains what AI can do. Models commoditize. Ontologies compound. Give the Totogi Ontology a test drive to find out what I’m talking about.
SoftBank just took out a $40 billion unsecured bridge loan—its largest dollar-denominated borrowing ever—and S&P downgraded the company’s credit outlook to negative the same week. The loan matures in 12 months, and repayment hinges on OpenAI going public before March 2027—a company that won’t be profitable until 2030. Meanwhile, SoftBank is also spending money training its own telco-specific LLM (see above). If you believe frontier models are so valuable that you’ll borrow $40B to own 13% of one, why are you simultaneously building a vertical model that the same frontier will absorb within a year? Pick a lane, Masa. Either the moat is the model, or it isn’t. (Hint: it isn't.)
AI is collapsing the cost of building software. It was already obvious, but Benedict Evans’ pithy, 16-minute keynote at MWC reinforced it. When AI can generate, integrate, and deploy code at a fraction of the old cost, the $50 billion+ professional services industry built on BSS complexity starts to look structurally unnecessary. Lower software costs don’t just compress margins; they eliminate the dependency/consulting model those margins are built on. Remember what Totogi built in Düsseldorf in one day: 617,000 lines of production BSS code, six core modules, in nine hours from one engineer. The old world is ending. If you want to launch yourself into the new one, call me. I can help. 🚀
AWS is partnering with Ericsson, Amdocs, and Nokia to move AI directly into the telco execution layer: network optimization, workflow automation, and agentic operations. These new tools signal a shift for AWS from selling raw infrastructure to providing a full stack that includes orchestration and AI frameworks, aiming to compress telco deployment cycles from years to months. It’s a powerful move. But here’s what AWS doesn’t have: a representation of how your specific business actually works. AWS can run your operations. It can’t understand your billing logic, your provisioning rules, or why “subscriber” means five different things across your systems. That semantic layer is the missing piece—and it’s exactly what an ontology like the Totogi Ontology provides. AWS plus ontology is the real architecture. Use both, and you’ll be cooking!
Deutsche Telekom is testing an AI assistant embedded directly into live phone calls—triggered by voice, no app required. I like the initiative here, moving AI into the product layer where it can be a service differentiator that subscribers will actually experience. Cute idea! If it works and is super awesome, great. If operators can make AI part of the core offering, they’ll have something to compete on beyond price and coverage maps. But let’s not just throw AI spaghetti at the wall to see if it sticks. Innovation is important. Still, customers want things that work. If this feature is clunky, people will be annoyed, and it’ll end up detracting from the brand. I see you trying, DT! But the bar is high. This better be amaze amaze amaze!
Network APIs for location, identity, and quality of service keep getting pitched as the telco industry’s next growth engine. Will they finally deliver? McKinsey sized the opportunity at $300 billion. Aduna launched with 11 Tier-1 operators to create a global API marketplace. Nokia’s Network as Code platform has grown to 50+ partners. And yet, most deployed APIs today still focus on simple functions like SIM-swap detection rather than advanced, revenue-generating interfaces. Enterprise appetite is real, but commercial traction is slow. And the only way to realize significant revenue is to get enterprises to use it. Until operators treat APIs as a product business with real P&L accountability instead of an IT side project, the monetization stays theoretical.
Ferry Grijpink, the McKinsey senior partner who co-authored the $300 billion network API forecast and shaped how a generation of telco executives thinks about 5G monetization, just left the firm after 22 years to join Verizon. This is worth paying attention to. When the person who wrote the strategy deck goes in-house, it usually means the operator is done planning and ready to execute. Verizon has been investing heavily in API exposure through Aduna and its own developer platforms. Hiring the architect of the industry’s most-cited growth thesis signals Verizon thinks it’s time to take that $300B slide and make it a reality. Whether Grijpink can move faster inside one operator than he could advising dozens is the bet—but Verizon just hired the guy who literally defined the opportunity. Here’s to wishing Ferry the best of luck in his new role! 🧣🥂🥳
