Did EchoStar force SpaceX’s hand to secure that $17 billion spectrum sale? Industry sources suggest EchoStar used its $1.3 billion satellite contract with MDA Space as leverage to force SpaceX into paying top dollar for spectrum the FCC was investigating for underutilization—then immediately canceled the MDA deal after closing with SpaceX. While Musk accused EchoStar of spectrum hoarding, maybe Charlie Ergen’s poker-playing background taught him something about building negotiating leverage that Silicon Valley bros never learned. EchoStar walks away with $17 billion to pay down crushing debt while SpaceX gets spectrum it desperately needed for Starlink expansion. Everyone’s analyzing what SpaceX gets out of this deal, but Ergen might be the actual winner here—sometimes the best business strategy is knowing exactly when to fold your cards and cash out.
AT&T’s Yigal Elbaz claims OpenRAN needs “scale from established operators” to survive, but he’s missing the real lesson from DISH’s shutdown. DISH’s cloud-native OpenRAN network worked flawlessly at 20% of traditional infrastructure costs. The problem wasn’t technical; it was commercial. No customers, no revenue, game over. While AT&T burns $14 billion on Ericsson’s “open” RAN, DISH proved you could build modern telecom infrastructure with 80% cost savings through true vendor disaggregation and public cloud architecture. EchoStar’s $23 billion spectrum fire sale to AT&T isn’t an indictment of OpenRAN technology—it’s proof that even the most efficient network architecture can’t save you from a broken go-to-market strategy. The takeaway for MNO execs? DISH solved the hard technical problem but failed Business 101, leaving a perfect blueprint for operators smart enough to combine cloud-native efficiency with actual customers.
SK Telecom struck a partnership with OpenAI, giving subscribers two free months of ChatGPT Plus, adding it to the operator’s already stacked AI roster alongside Anthropic, Perplexity, and NVIDIA partnerships for everything from telco-optimized LLMs to GPUaaS offerings. While most operators are still debating whether AI is hype or reality, SKT is building the ecosystem that will define next-generation telecom services in Korea’s emerging “full-stack AI nation.” This isn’t about offering AI as a side dish. It’s repositioning the entire telco value proposition around intelligence rather than connectivity. The contrast is brutal: European and American operators are still figuring out their AI “strategy” while SKT is monetizing AI partnerships and training customers to expect intelligent services as standard telecom offerings. Watch this playbook closely—following SKT’s AI ecosystem experiments might give you a head start on what works when you adapt it for your own market.
AT&T is testing an AI-powered digital receptionist that automatically screens and manages incoming calls—including identifying spam and fraud. Using multiple LLMs, the system can autonomously take messages and schedule delivery windows, plus create real-time transcripts. This network-based receptionist won’t drain your battery and works even when phones are out of cellular range. (Meanwhile, Apple just launched call screening in iOS 26, where unknown callers get automatically screened for their name and reason for calling before your phone rings.) AT&T aims to reduce costs by $2 billion through AI innovations like this. I love seeing both approaches, but consumers will probably prefer Apple’s: it’s included, works everywhere, and gives users control instead of being another carrier upsell. The downside of AT&T’s network-based solution is it likely won’t work when you’re traveling on other carriers’ networks—exactly when you need spam protection most. The lesson: Carriers should focus their AI investments on features handset OS vendors can’t replicate—like helping subscribers optimize their plans, predicting network usage patterns, and growing ARPU through intelligent service recommendations, not competing with software that Apple can build into iOS easily and distribute instantly.
Oracle reportedly signed a $300 billion, five-year cloud contract with OpenAI—the biggest cloud deal in history—and it’s definitely not because Oracle Cloud Infrastructure suddenly got awesome. This is OpenAI diversifying away from Microsoft Azure after realizing that depending on your biggest AI competitor for infrastructure is strategic suicide. Microsoft’s massive OpenAI investment looked like a partnership until Redmond began developing its own competing models and using OpenAI’s technology to build products that directly compete with OpenAI’s offerings, making Azure reliance increasingly awkward. OpenAI already uses Oracle’s compute services and signed that Google Cloud deal, proving it’ll pay premium prices to avoid vendor lock-in with Microsoft. I guess Oracle became relevant not through innovation, but because it isn’t one of the big three hyperscalers competing directly with OpenAI in the AI market. Talk about being in the right place at the right time.
Google announcing a £5 billion investment—during Trump’s state visit while facing antitrust lawsuits back home—isn’t coincidence. It’s strategic infrastructure play. The tech giant is opening a data center in Waltham Cross, expanding DeepMind operations, and promising to create 8,250 jobs annually while committing to train 7.5 million people in AI skills by 2030. The partnership with Shell for renewable energy supply shows Google is constructing vertically integrated infrastructure that makes its GPU-as-a-service (GPUaaS) offerings a one-stop shop. For MNO executives pushing their own GPUaaS strategies, this is the harsh reality: while you’re partnering with third parties for compute and power, Google owns the entire stack from renewable generation to AI inference, making its cloud economics unbeatable and your GPU aspirations look increasingly fragile. Ouch.
Handset brand Nothing raised $200 million at a $1.3 billion valuation to build AI-native devices, and here’s why it’ll succeed where Humane spectacularly failed: Nothing isn’t trying to replace your smartphone. While Humane burned $230 million on a $699 wearable that overheated and lasted 2-4 hours, Nothing has already shipped 5.1 million working devices and crossed $1 billion in sales by focusing on software innovation within proven hardware form factors. Humane tried to solve a problem nobody had—getting rid of phones entirely—while Nothing’s betting on evolutionary AI integration through “hyper-personalized, context-aware operating systems” that enhance rather than replace the smartphone experience. I’d say the takeaway for telecom executives is to keep close tabs on these AI device experiments to spot which ones consumers actually embrace, then be ready to pivot your partnerships and services strategy before your competitors figure out what’s working.
