Totogi’s official press release summed up its live AI coding demonstration at TelecomTV’s AI-Native Telco Forum: a single forward deployed engineer generated 617,000 lines of production-quality BSS code across six modules in one day using BSS Magic. The platform uses a telco ontology built on TM Forum Open Digital Architecture and 3GPP standards to give AI the domain expertise needed for production code—the reason it can produce such fast and high-quality software. This is the kind of speed that should terrify legacy BSS vendors. When you can build in days what used to take years and millions of dollars, the entire vendor relationship changes. Next step: connect BSS Magic to existing systems and prove we can integrate with closed legacy platforms without vendor cooperation. (I’ll let you in on a little secret: we can do that, too. See this week’s blog for details.)
Totogi’s live AI coding demonstration set a new bar for AI coding in telco and generated massive industry coverage. TelecomTV ran two feature articles, Telecoms.com wrote about it, TheFastMode covered it, and the story was syndicated internationally. The consistent headline: “Totogi generates 500,000 lines of production BSS code in one day—live.” Edward Finegold said Totogi “stuck their necks out” to show what’s possible with a new approach to BSS. What made the demo newsworthy wasn’t just the technical feat (which was huge, by the way)—it was proving that legacy BSS vendors’ “years and tens of millions” timeline is a business model, not a technical limitation. And it’s time for that business model to DIE. Even skeptical coverage reinforced the core message: AI-native architecture built on a telco ontology is fundamentally different from retrofitted AI bolted onto decades-old platforms. Does this threaten Amdocs’ throne in telco? We think so… 🤔
NEC is acquiring CSG Systems for $2.9 billion to expand its SaaS BSS portfolio, adding CSG’s customer care and billing services to its Netcracker subsidiary. With CSG’s revenue at $1.2 billion, this valuation is 2.4x revenue—a consulting company multiple, not a software multiple. Compare that to true cloud-native SaaS companies trading at 10x revenue. CSG gets valued like services because every Tier 1 customer demands custom versions instead of using multi-tenant SaaS. Legacy BSS vendors take 18-24 months to deploy new features while AI-native platforms like Totogi generate production code in days. NEC is buying market share in a shrinking category: BSS that requires armies of humans to maintain. When AI can write, test, and deploy charging logic faster than your integration team can schedule meetings, consolidation becomes a bet that scale matters more than speed. It doesn’t.
The industry is completely split on what “AI-native” actually means and how long it’ll take to get there, according to discussions at TelecomTV’s AI-Native Telco Forum. Docomo sees 6G rollout in 2030 as the opportunity, IBM predicts 10-15 years due to legacy technology debt, and Deutsche Telekom claims telcos are partially there already. Here’s what I think AI-native actually means: redesigning every job and process with an AI-first mindset—not bolting AI onto existing workflows. The transformation will take 10-15 years, but AI accelerates and improves your business from day one as you replace human work in network ops, customer service, and field operations. Stop debating definitions and start redesigning processes now. Every quarter spent arguing about what AI-native means is another quarter you’re paying humans to do work that AI can do better today.
Omdia predicts telco spending on cloud tech and software will jump from $17.4 billion in 2025 to $24.8 billion by 2030, showing 7.3% annual growth driven by AI implementations in network operations. Over 62% of operators now consider AI/ML support critical to cloud infrastructure decisions. Here’s the depressing part: public cloud usage for network workloads will only reach 13% by 2030—meaning 87% of that $24.8 billion is still going to private cloud infrastructure. Ugh! I’ve been telling telcos for years to move to the public cloud, and you didn’t listen. Now AI has arrived and you’re building “AI-ready infrastructure” in your private clouds while hyperscalers already have it running at scale—with the latest AI chips, economies-of-scale pricing, and ready-to-use AI services. You’re years behind. Jump on the public cloud to skip ahead and get all the benefits you need today. Remember, private cloud is still #fakecloud.
AWS is investing $5 billion in South Korea by 2031 to build AI data centers on the outskirts of Seoul, part of a broader $40 billion commitment across 14 Asia-Pacific countries through 2028. This follows AWS’s $4 billion investment announced in June with SK Group for a data center in Ulsan. OpenAI is also building data centers there with Samsung and SK Hynix, making South Korea the APAC AI battleground. Before you embark on your GPUaaS strategy, make sure you have hyperscaler bank accounts. These chips evolve every year, making that five-year depreciation plan obsolete before you even deploy. AWS can afford to replace entire data center fleets as new GPU generations ship. Can you? When your H100s are outdated and customers want H200s or whatever NVIDIA ships next, you’re stuck with stranded assets while hyperscalers write it off as the cost of staying competitive. Ouch.
AWS experienced a massive outage on October 20 that took down Netflix, Snapchat, United Airlines, and roughly 1,000 other platforms when a DNS race condition in DynamoDB created an empty DNS record. The three-hour failure cascaded across 113 AWS services and originated in US-EAST-1. Here’s what telcos need to understand: US-EAST-1 is AWS’s oldest, largest region and the launch pad for almost every new feature and hardware generation—meaning it has constant change and higher outage risk. AWS gives you cheaper pricing there because you’re accepting that risk, and it offers 69 additional regions across 39 countries so you can architect around it. For telcos moving carrier-grade workloads: design multi-availability-zone redundancy or run mission-critical services in multiple regions. As I blogged about years ago, the outage isn’t AWS’s problem. It’s yours.
Apple may finally strike a deal with SpaceX after rejecting Elon Musk’s partnership offer back in 2022 (reportedly because Tim Cook didn’t want to upset the carriers). But times have changed. SpaceX has acquired $17 billion worth of spectrum from EchoStar for direct-to-cell satellite connectivity and T-Mobile already has iPhones testing Starlink integration with iOS 18.3. Now that customers can get seamless satellite backup for coverage gaps, why pay premium prices for nationwide cellular networks? This is exactly what I’ve been warning you about. (See: The Wolf of MWC.) Musk’s endgame isn’t just Starlink connectivity. It’s building the infrastructure for interplanetary communication that happens to work perfectly on Earth too. Remember, on the way to Mars, telco is just collateral damage.
