#122 Telstra's new Connected Future plan | Corning's Kevin Hussey talks data centres

Music. Welcome to Comms Day Live, I'm Graham Lynch. This week we're doing a deep dive into a hot area in telecoms right now.

Viber connects into data centres which are being fuelled by the AI revolution. We'll be having an extensive chat with Kevin Hussey, an American-based VP for Corning, later in the show. But first, Australia's biggest telco, Telstra, is diving without reservation into an AI future. CEO Vicki Brady unveiled a five-year plan last week called Connected Future 30. And it's a lot more than just a rebadged version of cost-cutting programs such as T22 or T25 before it.

It's a complete reinvention of purpose and business model. Let's hear it from the CEO herself. We are again at one of those inflection points where technology, things like AI, is transforming. And so too is connectivity. We know as we step back in this moment in time and look at some of the key trends, firstly, we can see customer needs are changing. They're looking for connectivity to be more sophisticated and more flexible.

We know we have different attributes that our network can provide from uplink to reliability, latency, security. and we know these things are important for different segments of customers. We already have customers wanting to engage with us in different ways, for example, through APIs. How we anticipate and meet those changing needs of our customers will be crucial to our success. We also know that there is unrelenting demand for more data and more connectivity.

We're seeing that fuel a super cycle of investment in digital infrastructure. There is no version of the future that does not rely on technology and it all needs to be connected. As we look ahead over the next five years, we believe the pace and scale of that change will only accelerate. And that is going to drive more and more demand for connectivity. In moments like this, we think it's really important to be clear on who we are and what we do. We have significant competitive advantage.

And we see a massive opportunity ahead. And so do others. The environment that we're competing in is fast changing. So we must move fast too. Back in March at Mobile World Congress, there was a lot of optimism in the telco industry about the role we can play in this future. But there was also recognition that we need to think differently. Think differently so that we can capture some of the value that is created. That's essential to sustain our ability to keep investing to support it.

Our connectivity business has a long history of innovation. And so we see this as a massive opportunity for us. Our core connectivity business is strong, particularly our mobiles business. We have a successful digital infrastructure business with exceptional assets. And our network innovation and customer reach is unmatched here in Australia. Now, some of you have seen these moments before and might be thinking, why is it different this time around?

Telcos have said this before. So why will it be different? What is fundamentally different this time is the core of our business is transforming, and that's our network. It's becoming more virtualized, more software defined, and more autonomous. That gives us the ability to do things differently, gives us the chance to challenge the commercial models, to create differentiated offerings that meet different customer needs.

We're also seeing globally as telcos come together and align around standardized APIs. We have the conditions to disrupt the status quo and create new value. Again, being clear on who we are, our strengths and what our customers need will help us compete and be successful. Our Connected Future 30 strategy will see us double down on connectivity and radically innovate in the core of our business. We will do what we do best, and that is adapt and lead to shape the future of connectivity.

We will push the boundaries on our network leadership and make it simple for customers to get the connectivity they need in a fast-changing environment. We will step up our focus on being efficient and competitive, and we will reinvent commercial models so we can continue to invest for customers and communities. Our overarching ambition is to be the number one choice for connectivity in Australia. To achieve that ambition, we know we need to understand our customers even more deeply.

And we need to treat our network as a product with its own value. So we are thinking about our business differently across three layers and pushing ourselves on the opportunities in each layer to generate more value for our customers, be more efficient and more competitive, and in turn, deliver more value into our business. So let me talk through the three layers, which have a goal each.

So if I start with our customer layer called customer engagement, our goal is to lead in how we anticipate and deliver on the connectivity needs of our customers. Our second layer, our network layer, network as a product. To build and operate Australia's leading network and reinvent how we capture value from it. Our infrastructure layer, digital infrastructure, to be Australia's leading digital infrastructure provider.

Now, while most of the headlines around Connected Future 30 focused on investments and AI, I want to talk about something deeper, language. Because Telstra is not just changing what it does, it's changing how it describes itself. And that does matter. The company is now speaking in terms of layers, customer engagement, network as a product, digital infrastructure. It no longer talks about towers and ducts. It talks about ecosystems that stretch from wet to dry to sky.

The network is not just about national coverage. It's now a programmable platform. What's happening here is linguistic relativity in action. Telstra is reframing how we even talk and think about infrastructure, moving from the logic of utility to the logic of platforms. That shift is not cosmetic. It actually will drive how value is created, priced, and capitalized in the future. CFO Michael Ackland put it quite plainly. He said, we're going to need new economic models.

He's talking about things like monetizing latency, uplink, security, terms that come straight out of the cloud world, not the telco world. Telstra's language matches that world now. Vicky Brady says that AI thinking has been transformed by visits to the US. They now speak of agentic AI, API first. These are not the words of a legacy telco. It's not just speech writing, it's strategy. And it's backed by action. They're hiring cloud-native leadership, Stephen Worrell from Microsoft, for instance.

But here's the rub. While Telstra is changing how it's understood, our industry frameworks still cling to old paradigms. Arbitrary speed-tier flaws, 2D coverage maps, so on and so forth.

If Australia wants a future-proof digital economy, we all need to catch up. Music. Now, straight on to that very digital infrastructure that Telstra has been talking about. Corning, one of the world's preeminent fibre makers, and a really hot market for them right now, is in the data centre interconnect area.

I caught up with Corning's global vice president of data centres, Kevin Hussey, to talk about the explosion in connectivity requirements in the sector and where Australia fits in in a global context. I started off by asking Kevin, what are CEOs in the data center sector telling him right now? It's one great to be with you, but we are the number one challenge we're trying to solve now as these new GPUs come to market is really the density challenge.

So the way we tend to discuss with most of these CEOs is there's a supply chain challenge given the demand in the market. There's a density challenge they're trying to solve, which is largely creating problem number one. And they're looking for very innovative ways to solve some of the labor challenges that they're facing.

The unique thing that we're seeing happening is because of some of the challenges around power and labor and others, a lot of these CEOs are looking to build data centers in places they've never, they haven't traditionally built. And that creates a separate set of challenges. So power tends to be the one that leads the way, but that very quickly turns into a labor force issue that often very quickly turns into a speed of deployment issue.

So what we've, you know, our secret sauce at Corning has been to partner very closely with these companies and figure out the biggest problem they've got and tackle it with kind of our innovation machine. And that's where we spend our money. That's where we spend our time and creating this set of solutions that create this building block for the entire industry around the world to grow from. Okay. Now, Australia is already established as a high investment destination for data centers.

How's the view of this market from where you are? Yeah, I mean, look, it's got a lot of things going for it. Obviously, I mean, APAC as a region has a very resilient supply chain today. It's a very desirable place to make. It's a very desirable place for us. I think it's a very stable place geopolitically. In today's day and age that matters quite a bit, but it seems very ripe for infrastructure investment as well.

And so one of the things we're talking to a lot of our customers in the region, in particular in Australia, about is thinking about as suture ready as we can about the infrastructure that's required in the long-haul network, in the middle-mile network, and ultimately on these campuses.

And we're seeing it in different parts of the world. Again, these traditional hubs for data centers, where we've traditionally placed them, aren't necessarily the best places or the most desirable places for these large GPU clusters.

You know, what started as hundreds of GPUs to thousands to hundreds of thousands to now, conversations we're having around the world about clusters of a million GPUs or more is really where that's where the problem solving is coming in for both the location of these data centers, the labor force to build it, and the products to tackle the challenges that those create.

Okay, now you just mentioned GPUs there. That's interesting because NVIDIA's dominance in that space is receding. There's a lot more competition. What does that mean for data centers? Yeah, I mean, there's one, a lot of the hyperscalers obviously have created their own chipsets too. We're seeing the emergence of others like AMD and some of the other providers. For us, it has really highlighted the importance of having good foundational building blocks for these networks.

And, you know, we've focused in really the last five years on an entire new set of fibers, of connectors, of cables, and then ultimately of hardware that regardless of the GPU choice that operators in Australia or other places might make, provide really good building blocks for them to build an open network architecture that scales very nicely depending on the size of deployment.

So whether you're an enterprise doing your first on-prem GPU deployment, a colo or a multi-tenant data center player, or a hyperscaler looking to expand in the region or in Australia specifically, there's a solution set there that really enables them to grow nicely based on the use case that they're solving for. Okay, now you're an advocate for non-proprietary fiber solutions. Tell me about your message there and why you think that approach is relevant now.

Yeah, I mean, look, certainly in the last six months, supply chain resilience is never probably more to the forefront. So the ability for us in the industry, really, to manufacture locally and manufacture a fairly consistent product set for our customers is really important. And those things, by the ability to do that and do that at scale, it impacts cost. It certainly impacts, you know, having interoperable components really enables resilience.

Uptime on these networks has never been more important. And certainly, you know, building products that are very sustainable, the smaller, you know, as we miniaturize these things, you're taking up less space in racks, less space, less space in ducts, and ultimately shrinking the footprint that these products are taking without sacrificing anything in terms of performance and in fact, increasing the density of it.

And so that's where a lot of our capital investments have gone is in creating a product set like that, that regardless of location, regardless of technology, gives you those options. Okay, so if you've got, say, an open and non-proprietary fiber strategy, and you map that with open compute and open network architectures, obviously there's a symmetry and philosophy there, but what are the actual advantages of combining those three elements with a mindset towards being open?

Yeah, I mean, most of our conversations now, Graham, with customers around the world, latency is becoming a bigger piece of the discussion. And so having an open platform and products that enable them to almost spread the distance between these GPU clusters. So, you know, one of the interesting things that we're, one of the challenges we're really focused on solving now with our customers is Sydney and Melbourne as an example.

As these data center clusters continue to grow and expand outside of that, there are ways with new fiber technologies, for example, to get them to operate as a single brain. And so you can expand the distance that these clusters can be apart while they're still operating as one single unit, which provides a lot of advantages and flexibility as they're deploying new GPUs to expand their compute capability.

And so for us, giving them the ability to leverage disaggregated infrastructure models, like you said, and enable these ultra high density GPU clusters. I mean, like I said, designing a million GPU cluster today is not something even two years ago that was on our mind. And the work we're doing today with customers all around the globe to enable ones that are even bigger, clusters that are even larger than that, that operate as one, is really where we're spending our time.

And I think having a product set that enables that sort of open network architecture is actually absolutely critical for the industry. And again, it comes back to the cables, the connectors, and the fibers that you're inventing every single year now on this product cycle to keep up with the GPU cycle. Yeah, okay. Now, in the broader telecom market, we hear a lot about concerns about Fender lock-in, Fender monopoly.

Obviously, that's a big issue in network infrastructure, particularly with certain companies being banned from supplying the operator market. Is that phenomenon equally an issue in the data center area? Is it a danger of vendor locking as we see with the telecom operation?

Yeah, we saw it first with connectors, I think. I think there was a high sensitivity around having standards for connectors in the market. And so, you know, when we launched the MNC connector and kind of put our development work behind that, I think that helped push the industry into that as a good choice, as one of the good options.

We're seeing other connector technologies take hold. And more importantly, seeing multiple vendors invest in the ability to manufacture those so that you don't have this vendor lock into any sort of proprietary connector. And for better or worse, then, that gives any of these operators options in terms of who they choose to supply, their fiber cable, and ultimately their optical infrastructure.

And, you know, the other interesting trend we're seeing, Graham, is at a high level, there's far more risk in under-provisioning for fiber than there is in over-provisioning. And I would say one of the phenomenons that I saw over the last two years was, I think we as an industry probably underappreciated how quickly the GPU development would happen and ultimately the impact that would have on the necessary infrastructure. And so by having these...

Open architectures and avoiding this vendor lock-in, you really give yourself more options to architect along the way based on the technology that you're deploying. And we're seeing hyperscalers and enterprises all around the world take pause at times and rethink the amount of fiber that they want to have and provision in a data center, understanding that they want to have room for growth.

And so we're trying to give them as many options as possible as an industry to do that, that'll enable that growth. Okay. I just wanted to come back to Australia. Do you see any kind of unique attributes or characteristics in Australia that might present challenges or issues for the development of our market in the future? And I personally have a couple of things in mind, like a relative lack of energy and water could perhaps become one issue. But what issues do you see from your vantage point?

I mean, I think power is Certainly top of the list. Energy is A1A as far as I can tell. The other big catch up I would say that I'm seeing around the globe is really in the data center interconnect. So the long haul networks. Again, I don't know that anyone appreciated the level of investment that will be required in the long haul, in the middle mile, and ultimately on the campus of that.

And so we're starting to see people, I would say in the last year or so, we did that deal with Lumen that we made public last summer, and we're kind of executing on that. I would say there are multiple customers in multiple regions of the world that are looking to build larger interconnect networks to connect countries, to connect cities, and to connect campuses of data centers with all the new technology that we've talked about, particularly around new fibers and cables specifically.

But that interconnect piece of the network, I think, could be and I think will be one of the areas that will require significant, I'll call it, catch-up investment.

The power investments tend to be multi-year investments regulations play a part in some of those power issues but power i mean it all comes down to power density labor availability and then ultimately like i said i think these interconnect networks have to catch up because the amount of traffic these new clusters are going to drive is is you know 15 20 times more than i think anyone probably would have expected three or four years ago yeah i would agree with that totally now

let's get the crystal ball out.

Let's look ahead five years to 2030. What do you expect we're going to see in terms of developments in data centers over that time? You just mentioned before about some of the incredible changes we've seen in the last two or three years. So what can we expect across the next five? You know, the big challenges we're currently tackling are on a couple fronts. Number one, there is a big piece of these data center networks that are still copper today.

The back end of these networks that are connecting all these GPUs are copper. We're working very closely with customers all around the world about how that network looks when it's essentially an all-optical network. And so a lot of what we're doing is investing ahead in a new solution set with a number of customers around the world to try to solve that as fiber goes much deeper into the rack and as these clusters get exponentially larger than they are today.

And like I said, you're going to see campuses that have 500,000 million GPUs. I think I saw a note after the NVIDIA earnings yesterday that, you know, there's something like a doubling of GPUs for every data center that they're now deploying versus where they were a year ago. And something like, you know, 70,000 GPUs are being deployed a week.

You know the rate of deployment has has grown and so a lot of what we're spending time on now is making the right investment decisions ahead in capacity to support it number one in the region, number two a lot of this grand goes to you also got to automate the manufacturing, technology and so these cables you know these connectors are getting much much smaller you're putting you were putting two fibers in a connector now you're putting 24 in a connector and the connectors as small as it's ever

been and we're talking about 36 and 48 fiber connectors right the ability to do that at scale in the globe to enable this in the region and the country is critical as these get bigger. And there's some new-to-the-world technologies we're working at with a lot of the big players.

We've heard a lot about co-packaged optics, new fibers to the world, holocore and multicore fibers kind of on everyone's mind that will really enable both that data center interconnect on campus and that metro long-haul piece of the network that will really enable, I think, a step functioning capability for all of our customers and certainly for the Australian data center operators that we're looking to invest alongside.

Okay, well, that's fantastic. Kevin Hussey, thank you so much for joining Comms Day Live. Great to see you, Graham.