# TE Connectivity at OCP Summit 2026 The Interview

> Source: <https://www.servethehome.com/te-connectivity-at-ocp-summit-2026-the-interview/>
> Published: 2026-11-16 07:00:21+00:00

At the 2026 OCP Global Summit, I had the chance to sit down with Brian Payne, Vice President of Product Management for the Digital Data Networks (DDN) business at TE Connectivity. Many of our readers will know Brian from his 26 years at Dell, but he had been at TE Connectivity for about a month when we spoke. TE’s booth was built around three interconnected system-level racks, an optical interconnect rack, a sidecar power rack, and a copper IT rack, so we walked through the high-speed connectivity, the power delivery, and why TE thinks its system-level approach matters as AI infrastructure scales. We previously published a short booth tour video on STHLabs that got a surprising amount of attention, so we figured we would follow it up with the interview we did while we were there.

As always, we lightly edited this for readability.

## Brian Payne and TE Connectivity at OCP

**Patrick Kennedy:** Brian, could you tell us a little bit about yourself?

**Brian Payne:** Sure. First of all, thanks. It’s great to see you again. I am vice president of product management here at TE Connectivity. I’ve got a long history in this industry. I spent 26 years at Dell.

**Patrick Kennedy:** Could you give us a little bit of background about TE Connectivity?

**Brian Payne:** TE Connectivity is a very large organization that manufactures and provides connectivity across multiple industries. But one of the most exciting things we’re doing right now is in what we call DDN, our Digital Data Networks division. We have experienced massive growth over the last three years, all building out the AI infrastructure that’s powering everything we’re doing today. It’s an exciting space to be in, with tremendous upside ahead of us.

**Patrick Kennedy:** As part of that, TE Connectivity is well known for its contributions to OCP.

**Brian Payne:** TE Connectivity has been a part of OCP since 2017, so very early days. And as you can imagine, look around us. OCP has just boomed in terms of growth. TE Connectivity is focused on power and connectivity across all domains in the AI infrastructure. We have made tremendous contributions, and there are some exciting ones we’re going to talk about today.

## 448G and Next-Generation Front-Panel Connectivity

**Patrick Kennedy:** You have some things in your hand that I think you are super excited about. Could you tell us what these are and show us a little bit?

**Brian Payne:** Absolutely. At TE Connectivity, we’re thinking about how we deliver power for the data center and how we connect it. With every chip innovation, more and more bandwidth is coming. How do you get that out of the compute tray, and how do you get it into the network? One of the challenges is getting to 448. There are a lot of thermal challenges and signal integrity challenges to balance, and there’s a push in the industry to find the right solution. I would say there are two competing models out there. There’s XPO, which is a large.

**Patrick Kennedy:** And just to be clear here, these are 3D-printed models.

**Brian Payne:** The reason I’ve got a 3D-printed model is that just Monday of this week (*at the October 2026 OCP Summit*), in a partnership with NVIDIA and Cisco, we announced our new standard for getting to 448. That’s what I have in my hand right here. You will see that it’s a similar form factor to QSFP today. One of the things we’re trying to do is balance signal integrity, power delivery, and form factor, and not over-index on one or the other. We believe we have a winner.

When you talk about developing concepts, that’s exactly what we have here. The transceiver slides in, and we’ve got a unique way to compress it so that power delivery and signal integrity are handled separately. So our signal integrity is really, really strong. The design is also made so that, as we see the power of the transceiver escalate, we’ll be able to do cold plate liquid cooling to pull heat out of the system.

**Patrick Kennedy:** When you say 448G, you’re talking about the SerDes and really the per-lane communication speed.

**Brian Payne:** Absolutely.

**Patrick Kennedy:** That’s a big deal when we start talking about things like picojoules per bit, or how much power it takes to move data. But there are also a number of enabling technologies beyond USFP that you have behind us. Maybe we should start with some of the copper over here.

## Copper Backplanes and Midplanes

**Brian Payne:** One of the huge areas of breakthrough in the industry has been copper backplane technology, and a combination of copper backplanes as well as midplanes. What we have here is an example of a midplane design, where a compute sled slides in along with power delivery. By going with this midplane design, you’re able to separate compute from GPU and then easily connect those together through the midplane. Then on the back side, we’ve got copper backplanes. In terms of an unlock for these rack-scale solutions, it is a highly technical, highly engineered solution that we have to produce in massive volumes.

TE is one of the only companies in the industry that’s able to do that. There are very few in the market, and we’re excited about that. We are focused from the chip all the way to the switch. We’ve got innovation in near-chip packaging. We’ve got innovation around optics. We’re thinking about front and back I/O connectivity and the cabling in between. We’ve got innovation and products across all of those that are being deployed today at massive scale.

**Patrick Kennedy:** One of the interesting things in AI deployments is just how many lanes of communication are happening in every single box in a rack. How do you make that, number one, not use too much power and not create too much heat, but then also make it reliable? I imagine that’s a lot of what TE Connectivity needs to work on.

**Brian Payne:** One of the things I’ve discovered, and it’s been a pleasant surprise coming into TE Connectivity, is the R&D on the front end. How do we get through the material science challenges and develop the parts necessary to unlock these massive increases in bandwidth, and then deal with all of the power that’s required to push these chips and push the data across the network?

## Optics and Co-Packaged Optics

**Patrick Kennedy:** There are a number of technologies today focused on extending the usefulness of copper, but the next generation is really optics. I think you have a demo of that over here.

**Brian Payne:** The main thing I would say here is that you can go back to 10 Gigabit Ethernet, when we said the end of copper was coming, and copper is still prevalent in the industry. That being said, optics is there and solving a very, very important challenge. This is part of how we’re pathfinding and investing ahead. Earlier this year, we acquired an optics company, RAM Photonics. We’re bringing that into the product portfolio, and we’re massively investing in optics, from co-packaged optics as well as co-packaged copper, to getting the optics out of the system. So we’ve got a number of investments and new capabilities emerging to support that.

**Patrick Kennedy:** Folks have been talking about co-packaged optics in the industry for what feels like forever. What’s different about what’s happening today versus 10 years ago, when folks were saying, “Oh, we need co-packaged optics immediately”?

**Brian Payne:** I think it is the physics challenge, and how far you can route copper as we increase the number of lanes that are necessary. There is a point in time where we will have to make the pivot. I realize you could have said that 10 or 12 years ago, but it’s becoming more and more real as we face these challenges. That’s why our investment and productization are picking up in that space. But it’s not stopping our investment in copper, because copper is still the state of the art. It’s still important to deliver the outcomes our customers expect. That’s where we’ve got a dual investment.

**Patrick Kennedy:** One of the things you’re showing in the booth is some of the innovation around co-packaged optics, with connectors that go into the chip package so that you can service them. That was something folks didn’t know how to do even five or six years ago, and it’s a new innovation to bring co-packaged optics to the mass market and actually deploy it.

**Brian Payne:** When these are deployed at scale, it’s not just about getting the signal across the line. It’s about how you think about the system-level design and the operation, and what this looks like at scale. So to your point, serviceability is a critical design vector that drives innovation. That’s part of the TE approach: how do we not just get the technology working, but do it in a way that will support the massive volumes this industry is driving?

## Power: The Sidecar Power Rack and Horizontal Busbar

**Patrick Kennedy:** This is a power shelf and a power rack. Why do we have a power rack, with racks focused on data connectivity on either side?

**Brian Payne:** To deliver the performance the industry is driving for, a massive amount of power is required to power the chips, and then to power the switches that connect them together. We simply don’t have the power density today to fit all of that in one rack, and that led to the innovation we see today. We’re starting to see sidecar designs like what we have here, which is a full rack of power distribution and cooling capabilities on one side.

Then on the back side, when we get around there, we’ve actually got a busbar that runs horizontally, which is unique in the industry. Nobody’s done that. We did it to solve a problem with one of our large customers, and to deliver power to the next rack, the IT rack, with the IT equipment, the compute sleds, and the switches. That’s a great example of us thinking whip to chip. How do we put together a practical solution that addresses the ever-growing thirst for more and more power?

**Patrick Kennedy:** One of the really interesting things there is that the density challenge is actually in the nodes, so that you can get the copper connectivity within a rack. It’s actually using power to solve a connectivity challenge. And one of the big innovations is moving that power out of the rack and figuring out how to get a lot of current back into it.

**Brian Payne:** Right. In this industry, we have talked about power constraints and half-full racks for years and years. We’ve talked about busbars, and none of that really took hold until the AI boom. That’s what led to the innovation where we are delivering liquid-cooled busbars that span the rack to deliver power across without having to amp up the current and the overall heat dissipation. It’s about finding that balance in power delivery, but doing it in an efficient way.

The next frontier is high voltage. We’ve got pathfinding concepts on the back side here where we’re pushing towards 800 volt. That’s what’s going to be necessary to get to a megawatt of power at the rack level. So how do we solve practically for the limitations of today, and also prepare ourselves for the future?

## 800V and Rack-to-Rack Power

**Patrick Kennedy:** One of the cool things about 800 volt is that it provides a more efficient power distribution path, which means your data center is more efficient. But there are cool things you’re doing as well. That 800 volt busbar back there has liquid cooling.

**Brian Payne:** Absolutely. And there are even things we can’t show you yet. Getting to 800 volt, number one, we’re thinking about how you get 800 volt all the way from the source. Every conversion has losses associated with it. So how do we bring it from the busways down into the rack, and then deliver it efficiently across the rack? To run this efficiently, liquid cooling to pull heat out is a critical element, and it’s something we’ve solved and have deployed across the industry.

**Patrick Kennedy:** You have another solution, which is a big, giant rack-to-rack cable. Can you tell me a little bit about that?

**Brian Payne:** Part of the issue is that once you get the power down to the IT equipment and you need to span multiple racks, you have to have an efficient way to get it rack to rack if a horizontal busbar isn’t the optimal choice. That’s why we’ve got prototype rack-to-rack 800 volt cables on the back of the system.

**Patrick Kennedy:** And that prototype is another example of how TE Connectivity is doing pathfinding with the OCP community and with customers to figure out how to solve these practical challenges.

**Brian Payne:** Our customer engagement starts in our customers’ data centers and in their offices, but also at places like OCP, where you can see we’re just covered up with people coming by. That inspires thought, new ideas, and future innovation. It’s super important to be a part of this community, to get those ideas in, show people what we’re thinking, and then iterate and evolve the designs going forward.

## Pathfinding and Manufacturing at Scale

**Patrick Kennedy:** Could you give us some sense of how TE Connectivity doesn’t just make products for standards, but actually contributes to the pathfinding behind a lot of what is used in OCP?

**Brian Payne:** Absolutely. I’ll take power as an example. We think whip to chip. We’re one of the few, if not the only, company focused on the entire power delivery of the AI infrastructure at the AI rack. That means we see the bottlenecks before they occur, and we’re able to project ahead. The industry has the ambition to get to a megawatt of power at the rack level, and that has a lot of hard challenges. Behind us here, you’ll see some of the concepts we’re working on for how we’re going to deliver power rack to rack in the future. We get feedback on those from customers in real time, and we’re able to course correct and be ready when the industry needs us to go to full mass production.

**Patrick Kennedy:** I would imagine there are a bunch of steps that go into that. It starts with what problem you’re trying to solve. But then there’s also how you make something manufacturable, that you can build for a large AI data center and for an entire industry that’s really building around OCP.

**Brian Payne:** Everything starts with the customer. We’re partnering with the entire ecosystem, but especially the hyperscalers building out these AI data centers at a breakneck pace. The chip cycle, everything has moved faster, and we’ve got to be way ahead of where things are going: partnering, figuring out the problems they need to solve, innovating, delivering concepts, getting feedback, and continuing to iterate until we have the design the customers need.

Then that’s when we’ve got to shift to, “Oh my gosh, there are thousands of units to build to roll out in mass production.” We’ve got global manufacturing capabilities, and we’re able to rise up and produce high-quality, mass-produced parts to power the AI infrastructure we all enjoy today. TE Connectivity has a large global manufacturing footprint. We have 10 manufacturing facilities around the world, from China to Southeast Asia to North America, all focused just on the DDN business to power what’s going on in this industry.

## What Is Driving TE Connectivity

**Patrick Kennedy:** Brian, I’ve known you for years. At the end of the day, what are the big things that are driving you?

**Brian Payne:** It all starts with our customers. What do they need us to do? They need us to be at the forefront of technology innovation, and they need us to be able to deliver to support their business. That means we have to be able to design high-quality products that solve the technical challenges we have. But just as important, we’ve got to be able to manufacture at scale. We’ve got that balanced investment where we can do that across power and connectivity to push the industry forward.

**Patrick Kennedy:** Awesome.

## Final Words

A big thank you to Brian for taking the time to walk us through TE Connectivity’s booth at OCP Summit 2026. The broader theme was that AI infrastructure is no longer just a chip problem.

Getting to 448G per lane, scaling copper while preparing for co-packaged optics, and moving rack power toward 800V and the megawatt-class rack all have to work together at the system level, and then be manufactured at scale.
