New Zealand’s 2degrees is proving that cloud-native charging can move at internet speed, thanks to software from Totogi (where I’m acting CEO). Totogi has delivered over 300 feature enhancements with zero downtime and supported 2degrees through the launch of its MVNO wholesale platform last August. 2degrees can now onboard new mobile virtual network operators (MVNOs) with their own charging instance in days rather than months, while adding AI-powered churn prediction and automated customer journey mapping. This is what a modern telco looks like!
Is NVIDIA using telcos as its global AI infrastructure buildout strategy? With 18 AI factory partnerships spanning all five continents—from Orange and Deutsche Telekom in Europe, to Singtel and Jio in Asia, TELUS in Canada, and Cassava Technologies in Africa—NVIDIA gets to compete with hyperscalers without spending billions on land, power, and datacenters. Meanwhile, telcos are doing the heavy lifting: buying expensive GPUs, building specialized facilities, and competing against AWS/Google/Microsoft which have massive economies of scale. McKinsey & Company estimates a $35-70 billion annual GPU-as-a-Service market by 2030, but guess who's selling the shovels in this gold rush? WE ARE. NVIDIA wins TWICE with this approach: it gets paid for the hardware AND avoids the infrastructure investment. Telcos get to play in a market dominated by hyperscalers AND take all the CapEx risk. The lesson for telcos? Before you write that GPU check, ask yourself: are you building the fourth hyperscaler and getting those economics, or are we just NVIDIA's unpaid infrastructure team?
Oracle just landed the deal of a lifetime: OpenAI’s $30 billion-per-year cloud contract for 4.5GW of capacity through its Stargate joint venture. Let’s be honest—Oracle got lucky. While AWS, Google, and Microsoft have been spending hundreds of billions building global data center empires, Oracle was mostly an afterthought in cloud infrastructure. Then OpenAI desperately needed to diversify away from the Big 3 hyperscalers, and Oracle happened to have available capacity when OpenAI came knocking. This $30 billion deal is nearly triple Oracle’s entire current cloud infrastructure revenue. OpenAI will be able to diversify its infrastructure risk while Oracle scrambles to actually deliver on this massive commitment. This will be interesting to watch. 👀
The brewing Microsoft-OpenAI relationship tensions raise questions about what happens if the partnership dissolves. Microsoft has built a diverse AI-model library beyond OpenAI (even MSFT worries about vendor lock in!), and the breakup might actually free OpenAI to explore more flexible infrastructure partnerships (it’s already getting cozy with Google Cloud and Oracle). For telcos watching this drama unfold, the real lesson is that you shouldn’t put all your eggs in one large language model (LLM) basket. Build your AI applications, agents, chatbots, etc. with flexibility built in to switch LLMs with a single keystroke.
The top 5 barriers to private network deployment include cost concerns compared to Wi-Fi, spectrum access challenges, skills gaps, device compatibility issues, and worries about rapidly evolving standards. Net-net: they’re too fricking hard for the customer to run! Enterprises get spooked by the complexity. I know everyone hopes this is a growth opportunity, but until it’s easy-peasy like AWS Outposts, this idea isn’t going to take off.
UK adults now spend more time on mobile phones than watching television sets for the first time ever. The shift is driven by younger audiences who spend nearly 5 hours daily on phones but less than 2 hours watching TV. Telcos have won the attention war, but your unlimited plans are stopping you from monetizing this growth. Meanwhile, the hyperscalers are getting rich off every minute of that 5-hour daily habit. Will we ever kill the idea of the postpaid unlimited plan? 🧐
BT and Vodafone executives are encouraging their workers to use AI tools, with both operators acknowledging they’re not exactly hyperscaler-fast but insisting they can move when it counts. BT now has 50 AI solutions in operation while Vodafone has democratized AI access across tens of thousands of employees. Here’s the thing telcos are getting wrong about AI: it’s not about “augmentation” or cost reduction—it’s about fundamentally redesigning work to make your human workforce more valuable and productive. Stop thinking about AI as a cost center and start thinking about it as a human capital multiplier. Use AI to handle the grunt work while your people do the thinking, strategy, and innovation. This is how you actually get ROI from AI.
Wimbledon’s wrapped, and Light Reading’s Iain Morris (a tennis fan like me) was kind enough to put out a telco article with a tennis angle. Let me just say: I’m here for it! Iain wrote a thoughtful piece about Wimbledon’s automation of line judges using Hawk-Eye technology, drawing parallels to telecom’s AI concerns about eliminating entry-level jobs that serve as training grounds for senior roles. I don’t share his worry about AI and job elimination. Humans used to do calculations by hand and fill stock orders with paper slips. No one misses that inefficiency. Instead, I’m long on AI: it’ll improve, and these mistakes will vanish. Sure, we lose exciting moments like McEnroe’s “You CANNOT be SERIOUS!”, but we still have Sabalenka’s snippy sore loser comments and Dmitrov retiring mid-match against Sinner (while up 2-0 in sets!). AI WILL eliminate jobs. The question is whether telcos lead the transition or get dragged through it. Plus (and you probably won’t like this either, Iain), I think the tennis umpire can be, and should be, automated with AI as well. 🎾 🎾 🎾
Let the AI talent war begin! This LinkedIn piece from recruiter Nathan Greenhut examines who’s winning and losing in 2025’s hiring battlefield. The search for AI talent is reshaping compensation packages across every industry, with telcos finding themselves competing against hyperscalers for the same AI engineers. Smart operators should realize they’re going to need to build AI talent themselves rather than hoping to entice anyone away from Google, Microsoft, and Meta at salary levels that would make CFOs weep (did you see that OpenAI’s median comp is $875K and Meta is offering $2M+ salaries with $100M signing bonuses?!? Sign me up!).
Looking for AI talent? The second cohort of Gauntlet AI is underway. Full disclosure: My kiddo is doing it. But I’m worried that once his eyes have been opened to how easy it is to build apps with AI, he won’t want to go back to university and do everything the old, slow way to finish his Computer Science degree. One thing’s for sure, he’ll finally stop rolling his eyes whenever I tell him AI is the future. Mom FTW! 💪
