For Cloud AI, it’s all about the chips

Telcos, you’re sitting on a goldmine of data. Now, cloud-based artificial intelligence—Cloud AI—stands ready to use it to transform your network, processes, and customer experiences. But you’ll need purpose-built chips from the Big Three hyperscalers to unlock its power. They deliver unparalleled speed, cost-effectiveness, and innovation that simply can't be replicated on-premise. The future is here and you need this game-changing technology.?I explain all in my latest blog .

Anyone can be an MVNO with Gigs

Innovative tech firm Gigs is using the public cloud and APIs to make it easy for enterprise software developers to create and sell their own voice and data plans—in less than 15 minutes! Listen on?Apple Podcasts , Spotify , or the?TelcoDR website .

It’s out! Listen to?my executive interview ?with Silverlingings’ Founder Steve Saunders, the first in his Cloud 9 executive interview series. I talk about what’s on the horizon for telcos moving to the public cloud and the acquisition strategy driving my billion-dollar Telco Transformation Fund. Don’t miss it!

The #AIWars are on, and Google is feeling the heat from Microsoft. Google promoted?a loooong list of generative artificial intelligence (GenAI) products , just in time for its annual I/O developer conference. It highlights partnerships with other companies to use AI to?innovate, including French telco Orange, which is building a “next-generation contact center” that uses GenAI to transcribe calls, summarize exchanges between customers and service reps, and suggest follow-up actions. Contact center is the NO-BRAINER use case for GenAI, so if you’re not doing that, you should. Get on it.

While researching this week’s blog, I stumbled on this?estimate of how much it costs to run an LLM like ChatGPT , and a swag of how disruptive Microsoft’s move was to Google’s search perch: $30B. That’s a lot of market value up for grabs.?As I mentioned in the blog , Microsoft must be patting itself on the back about that, but it also needs to keep an eye on costs, as an estimate puts every ChatGPT query at about $0.36. Ouch, but worth it.

Adding “hyperautomation” (AI, ML, and robotic process automation) to your BSS ?amps up agility, scalability, and efficiency, so you can respond to market changes effectively and launch new products and services quickly. As the acting CEO of the?public coud-native BSS platform Totogi , I can tell you that you should be using Totogi to help break down your spaghetti-code BSS system (ahem, Amdocs) and rebuild it the right way in the public cloud. Keep on eye on Skyvera’s moves with the STL acquisition, as this is exactly what we plan to do with those assets.

Last week,?Totogi cut its charging costs by 20% platform-wide , thanks to?an insightful Twitter thread ?I found about how to lower Amazon Web Services (AWS) DynamoDB costs. I sent the thread to our engineering team and … BOOM! They made a small change that will drastically reduce our costs on the most expensive part of our stack—the database. This fast action is an example of the power of a purely public cloud design that a lot of people don’t understand: we can save money on our tech stack by making?software?changes. Can your private cloud, on-prem stack do that? No way, Jose.

Get inspired about the possibilities of network APIs and the public cloud by listening to?this AWS breakout session .?Leaders from AWS, Telefónica, Vodafone, and video technology company Cinfo talk about how telcos can provide better customer experiences by exposing 5G network capabilities to developers via network APIs, enabling the next generation of apps. TelcoDR recently bought CPaaS company Kandy and we are super excited about our plans for the technology. If you’re looking to replace Twilio at your enterprise customers, give me a call!

Fear of vendor lock-in is the main barrier to adopting the public cloud , according to a poll taken at TelecomTV's?recent Telcos and Public Cloud Summit. I can’t wait until this dead horse dies. You know what else you’re locked into? EVERYTHING! You’re locked into your chips. Your radio. Heck, even the database vendor you standardized on in the 1990s (that would be Oracle). Unless you plan to build EVERYTHING yourself, you’re going to have to get comfortable with standardizing on third-party software. With AI coming on so strong, you don’t have a choice. To use it at scale?you HAVE to use the public cloud . So go ahead already—move in and reap the benefits.

Still hoping to become a tech-co??Here’s a big reality check ?in the form of a chart listing the median total compensation at US tech companies for 2022. The numbers start at $270k and end up around $400k. What’s the median comp at US telco companies??I asked Chat GPT4 ?and the numbers are lower … a LOT lower. My guess was $75k, and I was HIGH (it’s about mid-$60k-ish). Net-net: it requires big money to hire strong tech talent. Make sure you’re ready to shell out the big bucks to build your tech-co team, or you’ll be hiring mediocre performers.?

If you’re following the news about companies leaving the public cloud to go back on-prem, there was never much of a story.?Charles Fitzgerald, on his Platformonomics blog, has introduced the?Repatriation Index? , which is a ratio of the combined revenues of Digital Realty Trust and Equinix, the two largest datacenter operators, to the revenue of AWS, the cloud leader. The index aims to measure the extent of workload repatriation from the cloud to the datacenter. The article reports that in Q1 2023, while the datacenter duo *did* grow faster than AWS for the first time in almost three years, it’s not a repatriation trend for a couple of reasons: the difference was minuscule (1.6%) and the gap in revenue between the two comparators is insanely large. There’s nothing to see here;?keep on moving your workloads to the public cloud!

Get this newsletter in your inbox!?Subscribe now .