Stargate. No, not that crappy movie from the 90s. I’m talking about the new AI joint venture funded by SoftBank, OpenAI, Oracle, and MGX (an investment firm) that intends to spend $500 billion building AI infrastructure in the US over the next four years. OF COURSE everyone’s flexing to show they have the big bucks to back it up, but regardless, it’s a huge amount of money. I guess it’s time for all us minions to start placing bets: will the money be made on training, or on inference? If it’s training, that’s good news for chip companies, like NVIDIA. If it’s inference, that’ll be good news for the hyperscalers. You know where my bet will be… ⛅️
Telefónica just installed a new CEO, Marc Murtra, after shareholders forced former leader José María Álvarez-Pallete to resign. With nearly 400 million connections and over €40 billion in revenue, I’m guessing shareholders wanted more than Álvarez-Pallete’s projected 1% growth rate. Maybe these guys should give Totogi a call since we made a money machine that generates 10% growth? Give it a try—it’s not hard to beat your 1% growth plan.
T-Mobile wants to use AI to reduce customer churn, and to do that, it’s building a whole new platform with OpenAI. I would love to be a fly on the wall when the operator hears the news that the new open source Chinese model DeepSeek can do what OpenAI can do for a fraction of the cost. Should’ve listened to TelcoDR before you signed that big $100 million AI commitment. Brutal.
Speaking of DeepSeek, it sure has upended the AI landscape with models rivaling OpenAI’s at a fraction of the cost, using just $6 million and open-sourcing its R1 model. The shockwaves hit hard: NVIDIA’s stock plunged up to 17% on Monday, erasing $600 billion in market value. For OpenAI, DeepSeek’s efficiency and open-source approach pose a major challenge, pressuring it to rethink pricing and strategy. Does its low cost mean the value will not be in the models, but instead at the application layer? Guess that’s great news for companies like Totogi!
Totogi is about to end the era of billion-dollar BSS failures with AI that delivers. At this year’s MWC in Barcelona, we’ll be showcasing production-ready, telco-trained AI software that works with your existing BSS systems and delivers immediate results. We've built our tech on top of TM Forum's ODA, and use advanced AI to connect to your existing vendors, reshape their rigid applications, and form an ODA-compatible foundation for innovation. Stop by the Totogi booth in Hall 2 or head to the AWS space to see it in action.
Mobile operator data is so valuable for AI that Rakuten CEO Hiroshi Mikitani says the company’s loss-making mobile unit is a critical component of its emerging AI strategy. Entering the saturated 5G market has brought the company more than $7 billion in losses since launch in April 2020—but it’s all worth it for the massive data volumes generated by Rakuten Mobile that will power specialized AI applications. (Have I mentioned telcos are sitting on a goldmine of customer data?) I bet Rakuten's regretting that private cloud strategy right about now, when it could have easily added public cloud AI software and accelerated its vision. Just sayin’.
AI in telco is still in its infancy, according to Vodafone’s Ignacio Garcia. And he would know, as the operator is already using generative AI (GenAI) to boost productivity, rewire operations, and launch new services. Vodafone has decided to partner with Google and Microsoft on infrastructure, and fine-tune models from Anthropic and Vertex rather than build its own tools from scratch. I can’t argue with that approach. But I think it needs to consider that both Google and Microsoft offer limited models, when it might want to try different models for different use cases (optimizing on speed, cost, and accuracy). This is not a one-size-fits-all problem.
Mobile virtual network operators (MVNOs) are better at customer satisfaction—at least in the UK, AGAIN. Four MVNOs made the list of Top 50 companies for customer satisfaction in the UK, as ranked by the Institute of Customer Service. The top performers were Tesco Mobile, Utility Warehouse, Sky Mobile and giffgaff (learn more about giffgaff’s approach on this Telco in 20 episode!). It’s not hard to see why: with no network to consume their attention, they focus exclusively on the customer experience. This isn’t a UK phenomenon; it happens in a lot of countries with MVNOs. What would happen if the telcos put NPS at the top of their list, instead of chasing the next G to install?
In the "DOH!" department, take a peek under the hood of BT Group’s in-house cloud platform in this report from Omdia (paywalled)—and then run far, far away. The telco plans to improve resource utilization and reduce fragmentation in its cloud environment, currently running on Canonical OpenStack for its mobile core. OpenStack is notorious for how difficult it is to manage and maintain. BT’s reliance on private-cloud, #fakecloud infrastructure limits scalability, creates inefficiencies, and slows down innovation. The public cloud offers everything BT needs to truly transform—global scale, advanced AI/ML tools, and a pay-as-you-go model that reduces costs and eliminates the need for fragmented CaaS implementations. If BT moved its workloads to the public cloud, it could accelerate service delivery, simplify operations, and future-proof its network for 5G and beyond. BT should pivot—sticking with private cloud risks falling behind in an industry that’s rapidly evolving, especially with the emergence of AI. The future of telecom is in the public cloud, and it’s time for BT to embrace it! ⛅️