As broadband providers prepare for higher-capacity networks and DOCSIS 4.0 deployments, network visibility has become increasingly important. Operators are looking beyond traditional troubleshooting methods and adopting AI and machine learning tools that can analyze network performance in real time. By continuously monitoring conditions such as signal levels, temperature and noise, these systems can identify anomalies and alert teams before service disruptions occur.
The next phase of network intelligence goes beyond faster repairs. By combining historical performance data with predictive analytics, operators can proactively address issues, improve customer satisfaction and optimize network reliability. As AI-powered applications continue to evolve, broadband providers will gain deeper visibility across the entire network, enabling a more comprehensive approach to maintaining service quality and supporting future growth.
Steve Saunders:
Hey, Todd, how are you?
Todd McCrum:
I'm doing well, Steve. How are you doing today?
Steve Saunders:
Oh, I'm better for talking to you because we're overdue for a chat. I mean, AOI ... First of all, for people that are not familiar with you, just give us the view from the satellite. What do you guys do?
Todd McCrum:
We have two primary businesses. One is the business that I run for AOI, the cable side of the business. And we supply HFC equipment to cable providers that are doing their upgrades readying for DOCSIS 4.0 and higher bandwidths.
And then the other side of our business, which has been in the news a lot lately, is the part that does optics for data centers.
Steve Saunders:
But the business isn't new, but it just seems like it's at a point where the market's coming to meet you as it were. Is that accurate?
Todd McCrum:
It's exciting to see the upgrades happening on both the downstream and upstream for the cable providers. We've always supplied a lot of optics. But when you see the number of hyperscalers that are just making massive investments in the space,
Steve Saunders:
What's at the core of your success?
Todd McCrum:
It's innovation on what we're talking about on the cable side of our business. When you talk about hardware, people have been doing this for years. I think what really changed the game this time is we did something that I've seen the industry trying to do for 40 plus years. And that is take intelligence and getting information all the way down to your last active in that network. And now we're taking that down in a very cost-effective way where we can get health information for every active between the home and the hubs for the cable industries.
Steve Saunders:
The idea of a hallucinating software agent on a network is going to fill a lot of my audience with dread. How do you de-risk that? I mean, or do you just not use LLMs, you just use machine learning? I mean, where do you set the line?
Todd McCrum:
When you look at it now and you think of how much information the machine can go through and we can help it with correlations. We give it parameters and say, "This is what we should see in terms of the noise floor. This is what our levels should be. This is what the temperature should be." The machine can send an alarm and say, "I've been monitoring this for 48 hours and it's where it is. My history of logs would tell me when this happens, something's not seated right and it's not dissipating heat. Or we have a power problem."
This type of ML and AI now allows to not have to take a plethora of analysts and can sort through this data and look for anomalies. And then the operator can decide what those parameters are and when to alarm on it. So now we move from getting a call and going directly to the trouble source and having faster time to repair. And now we move to preventative care. Now we start to get a lot of data that says historically when we see this, this is what the technician put in and said was the problem. And we build this database so that now AI can go through and say, "Hey, I'm seeing this and all of the previous data tells me that this is what we should be looking for." Send somebody out.
Steve Saunders:
Machine learning is obviously incredibly powerful, but also pattern recognition is hugely powerful for people in the industry which we're in.
Todd McCrum:
People are thinking, "Well, this is going to take jobs." And my answer to them is no, it's letting people do the job they were actually hired to do. Now as opposed to constantly doing post-issue work, now you're doing pre-issue preventative work. Now the customer sat numbers are going up and time to repair is getting improved. And now they're working on future looking things as opposed to in the rearview mirror on things that have already happened.
Steve Saunders:
Humans are absolutely the critical component of that network. But they're now being accessorized with distributed machine learning or AI of predictive capabilities.
Todd McCrum:
When you look at the stage we're at right now, what we've done is we've proven that we can bring all this data back to a central location. The additions that we launch in the coming months and leading up into the big SCT show is taking and putting applications on top of that data. And even in some cases, integrating this with some of the other people in our industry to take our data and the data they get from their solution sets, and being able to overlay that and now look at the entire network. And give a really good picture of everything that's going on in their network. And now it's not just sending someone out for a problem in the network. It's the proactive look at how do we improve the quality of service in the home?