That 10% revenue increase in 12 weeks that Totogi delivered for a Tier-1 operator is making news. It should! This solution from Totogi, where I’m acting CEO, achieved 5x the effectiveness of traditional customer value management (CVM) software—right out of the gate. And we think we can get it even higher. ICYMI, read all about Totogi’s “money machine” in my blog from January—then come see me at MWC25 to get it for your telco.
Rumor has it that Japanese tech firm NEC is considering a bid for CSG Systems, the U.S.-based OSS/BSS provider that counts Comcast and Charter Communications as customers. The deal, if it happens, would expand NEC’s footprint in the telco software space. Nobody’s saying how much the offer would be, but I can’t wait to see the price paid on that $1.2 billion of revenue. Currently, with a $1.6 billion market cap, it screams a services valuation, not a software valuation. Reminds me of that time Amdocs bought Openet (which was also a services valuation, not a software one). 🤔
DeepSeek-R1 is now available on AWS. In case you’ve been living under a rock, DeepSeek is the Chinese open-source large language model that rocked the markets recently with performance that rivals the top U.S. models. At first, it was reported that it cost only $6 million to train; now rumor has it that it may have cost $1.3 billion instead. OpenAI was so grumpy about the challenger that it accused DeepSeek of stealing its data, which is rich, given that a group of news outlets led by The New York Times is suing OpenAI for stealing its content. What goes around comes around, I guess?
Feeling the “deep” heat, OpenAI made a surprise release of a new o3-powered “Deep Research” mode for ChatGPT Pro users. The feature is designed to carry out extensive, web-based research, integrating various data sources (text, PDFs, images) into a cohesive analysis, and producing professional-quality reports—"accomplishing in tens of minutes what would take a human many hours.” These systems have achieved super-high-level quality at machine speed, as evidenced by Deep Research’s new record on the “Humanity’s Last Exam” benchmark, with 26.6% accuracy across 3,000 complex questions spanning 100 subjects. AI keeps getting better, and we’re still in early days. I'm so pumped!
While we’re on the subject, I loved this post from Aaron Levie on X about AI agents, and how they will coordinate with each other to accomplish complex tasks. It’s a great explanation of the coming “Agentic AI Era,” and also a nice intro to what you can do with BSS Magic! Has your org started to build cross-platform chained agents yet? Send me a DM and tell me all about it.
Speaking of agents, we used our own agentic platform BSS Magic to do a billing system migration at a Tier-1 North American telco in just TWO WEEKS—a process that traditionally takes 4-8 months! By using BSS Magic, we cut project hours by 90% while delivering ZERO downtime. Our secret? AI-powered ontological mapping that generated interface code and data transformations automatically. Be sure to stop by Hall 2, Booth 2G51, at MWC25 to learn how we can also help YOU modernize your BSS in days, not months. 🚀
Hyperscaler earnings reports are in, and though cloud growth has stalled for all three major cloud providers, they all have plans to spend billions on AI infrastructure. Alphabet is giving itself a $75 billion budget for AI investments in 2025. For Microsoft, it’s $80 billion, and AWS expects to fork out $100 billion. Charles Fitzgerald of Platformonomics shared a few quick takes on Microsoft, Alphabet, and AWS ahead of his full 2024 retrospective report. Check them out, and be sure to catch Chaz on a Telco in 20 podcast coming soon!
Gauntlet AI, an AI training program where participants work 100 hours a week for 12 weeks straight on nothing but AI software development, started with 220 participants. After five weeks, it’s down to 130. I’ve been mentoring Austen Allred (@austen on X) on how to run this program “Trilogy University” style, like I did back in the 1990s (where we made people do the 100-hour weeks, too). I gave a talk to the team last week about how you gotta work your ass off if you want to be successful in a leading edge field. I'm sure I scared the crap out of them, but it was a great talk. I may share it on my YouTube channel...
And last but not least, here are 321 real-world generative AI use cases from the world’s leading organizations! I came across this awesome list from Google Cloud, initially released last April and updated in December. It’s great to see telcos on the list, like… Telecom Italia, Bell Canada, TELUS, Verizon and Vodafone. Hooray! Who else is doing fantastic things with AI? Let me know!