- Ericsson exec Joe Constantine argues that trusted, accurate data—not better models—will determine which telcos win in the AI era
- Flattening network architecture and deploying a horizontal data layer will be key to achieving this
- Telcos are working on it, but face a number of constraints
There are a lot of things that keep Ericsson Americas CTSO Joe Constantine up at night. How do we extract more value from networks? How can we best deploy capital to reap multiples? How can we predict and prepare for where things are headed with AI in the mix? But perhaps the biggest problem on his mind has to do with data.
“The companies that will win and prevail are not the companies that have the best AI engineers or AI models. It’s the companies that have the best access to data and trusted data,” he told Fierce. “That’s the holy grail.”
It’s no secret that telecom networks generate reams of data every microsecond. That data comes from many different sources in different formats. Sometimes the data lacks integrity, sometimes pieces are missing, stuck in siloes or lost in translation. So, the big problem telcos are – and have been – struggling with is how to effectively capture, curate and authenticate all that data in a way that makes it usable.
Solve that, Constantine said, and then “applying AI is the easy part.”
The question then becomes how?
Constantine said the first step is creating a horizontal data layer that stretches across all facets of an operator’s network. Not just the RAN but also the transport and compute infrastructure. Sure, this can be done today, but not easily. Today, those that have solved this problem have done so with manual stitching together of these systems, Constantine said, but vendors need to start working together to make this easier and provide a path for upgrades.
Creating a layer of trusted data won’t just benefit operators on the AI front, it’ll also help with automation efforts, Constantine said. Sure, a handful of operators have demonstrated Level 4 automation today, but they’ve done so in vertical siloes. To achieve the kind of automation operators will need to dynamically handle AI traffic, automation needs to be applied horizontally.
This isn’t just an Ericsson perspective either. Constantine’s statements around the importance of managing data jibe with the results of a recent Cloudera survey, which found telcos are ahead of the game with data visibility and access but behind the curve on data governance.
Yes, data governance – the policies around how data can be used and by whom – is different from data fidelity, which refers to whether data is accurate and complete. But the message behind both is the same: operators need to get a grip on their data to move beyond networks based on quality of service to intent-based, quality of experience connectivity.
A work in progress
To be fair to operators, Constantine noted this is something they’re already very much thinking about. He said every CEO and CTO he’s met with across the board wants to create exactly the kind of system described.
“The top 10, 20 carriers on the planet, all of them are looking into this because data is monetizable,” he said. “If they want to win, they have to be able to make sense of all that information.”
But to get there, they have to battle a few constraints.
“It’s a blessing and a curse to be us,” Constantine continued, referring to the telco ecosystem as a whole. “We deal with critical infrastructure…We can’t be reckless in the way we do business. People always give me grief saying ‘Well, others, they do this and this,’ but very few of them have anything real time, [the way] that we do. They don’t have as much regulations as we do.”
He added the telecom industry has a whole trajectory of technology upgrades on its roadmap, but has to balance those ambitions with short-term monetary constraints.
Some operators, though, are forging ahead.
AT&T AVP and GenAI product lead BJ Nab recently told Fierce that the Big Data Center of Excellence it created more than a decade ago has morphed into its Data Office, and its experience in this realm has played a critical role in feeding the AI results it has seen to date.
And during Verizon’s Q1 2026 earnings call, new CEO Dan Schulman pointed to a data layer as a key component of its shiny new four-layer AI tech stack.
“We have a data and intelligence layer where we're…formatting all of our data – and we have a huge amount of data that's both structured and unstructured – putting it into the right formats, layering on top of that both LLMs and SLMs,” Schulman said.
All in all, Constantine said it is changes like these that will drive the next wave of telecom innovation.
“If you really want to cater and extract value out of that next generation connectivity, we have to flatten that architecture,” he concluded.