We've seen consecutive growth inside the company, and we've seen the ratio in business working closer to 50/50 between defense and commercial as we've grown in both. We've done that by establishing this position in global markets that are really, you're reading about every day, right? Moving artificial intelligence, sensor processing, and autonomy out on platforms and vehicles, quite large markets. I'm going to show you some data on that. For a small company, we're in a good financial position. In 2024, we saw consecutive growth in revenue quarter on quarter after moving on from some legacy contract manufacturing work we had done. We also saw year-on-year growth as we closed out Q4 of 2024. Strong balance sheet with little to no debt and enough cash in our balance sheet to sustain the growth that we're looking for going forward.
I'm going to go to this slide next, but this presentation is available on our corporate website, so definitely feel free to download it. Company, as I mentioned, a mix of commercial and defense. The reason you see more commercial in here is I'm going to take you through—we run in two business segments. We have a segment in the U.S. called OSS and a segment in Germany called Bressner. OSS is where we focus more on this edge compute capability. Bressner is more of a value-added reseller business that we operate out of Europe. You can see the key technology partners over here on the right, the ones you would expect to see for operating in data center-type technologies. What that reduces to for us, for edge applications, is the product set you see there on the bottom.
We will take the biggest, strongest supercomputing capability that you would find in a data center, and we can shrink it down and build it and ruggedize into something legitimately the size of two shoeboxes put end to end, and can provide not only the benefits of that compute, but extremely faster latencies to be able to move data around. As you can imagine, in this aspect, there's also a lot of data moving around, so we offer storage systems that coincide and work well with the processing power that we have. Then we do have some software capability, less at the application level, more down closer to the firmware and middleware layer that helps facilitate the efficient processing and control of these data center-class systems as they work out in the edge. This is an overview of those two businesses I spoke about.
OSS, headquartered in Southern California, just north of San Diego, and then our Bressner outfit, which is out of Munich, Germany. Again, the OSS segment is more about design, development, integration, build, test, and delivery of these edge processing rugged systems. Bressner is more of an integrator of existing vendor product lines and delivers those in widely commercial markets, more focused in Germany, the EU, and the Middle East. Examples of some of the customers that we deliver to there on the OSS side, not surprisingly, a mix of large defense contractors, end-use Department of Defense services, Army, Navy, Air Force, Marines, and then a number of commercial companies that operate in some of the segments I'll show you in some upcoming slides. Where we found our advantage or discrimination, clearly we ruggedize this commercial hardware.
In those big buildings, temperature is controlled, the buildings are strong and secure, they can withstand earthquakes and everything else, so there's not a lot of issues. You throw some simple stuff in a rack, that's not going to work when you put a couple of Marines in a tank or a combat vehicle, right? The stuff needs to be able to survive in those environments. Shock vibe, electromagnetic interface, cooling, salt, water. We deal with the ruggedization of being able to move these elements there. We do advanced cooling solutions. If you're following the data center world and the GPU processing, the wattage of those now are getting up to the point where you cannot cool the GPUs with fans anymore. You're having to use newer technologies. We are already using technology such as direct-to-chip liquid cooling, full immersion cooling.
You take a whole server processor, immerse it in a liquid, and we have delivered systems in those legs. I mentioned latency. We have developed some key partnerships, in particular with NVIDIA. We're a tier two OEM supplier, so we have special access in terms of capabilities of what we can do with their GPUs, how we can integrate them and use them and make the highest center of advancement for that. According to the technologies that are used in architectures today, you can see there across the bottom some of the advantages that we're bringing in terms of more memory, faster latency, huge advantages in computational performance. The biggest thing likely, though, is we bring that down in size, weight, and power.
The space that we operate compared to current technologies, and I'll talk about this in a little bit, is significantly less, and we can bring more power capability in that arena. I won't spend a lot of time here, but we did refocus the team and the ability to address more broadly defense programs. I added a new Vice President of Sales, we added a new CFO, and a new VP of Operations. Not only with experience in running defense programs and large value defense programs like $100 million plus programs, but the majority of us have also been part of large defense commercial portfolios in the $200 million-$700 million range. We know what big companies look like and how they need to run.
As we scale and grow the company, we've got that experience to help build from the lessons learned that we've already built. I started the company in 2023, where we started the strategy pivot to augment the commercial with defense. 2024 for us was about establishing a position for growth and driving pipeline and orders. We achieved that with some significant book-to-bill ratios in the order of 1.2 for 2024. That led to that growth. 2025 for us is about extending that growth and that pipeline and that win and book-to-bill ratio, but starting to focus on bringing profitability back up alongside that and growing the company steadily there.
2026 and beyond will represent a point where our profitability and all will continue to support reinvestment in that, and we'll look to greater expansion beyond the U.S., likely international, and be able to start to look at M&A opportunities. Where we're positioned, our products, as I mentioned, on the defense battlefield space, air, land, sea, airspace. One of the nice things about our company is we'll design a product, and it's not bespoke to a certain market or application. It's actually almost agnostic. Example, one of the servers that we build today, we build and deliver to a U.K. Navy for their submarine for sonar processing. It was the exact same box we sold to a commercial company who was doing autonomy for autonomous long-haul trucking. So we do have the benefit of designing elements and systems one time and then being agnostic and applicable almost across any market.
On the commercial side, where you might see us, we're doing some things in some interesting niches in the data center, especially where people are focused on high density of GPUs in a small, tight space. Anything industrial around robotics or application of AI or processing in an industrial or manufacturing environment. IoT, smart cities, and autonomy. This is where we kind of see that broad collaboration and connection of autonomy and AI and sensor processing in and around moving things in autonomous fashion throughout cities and around those environments. Where you might see us, if you think about an environment where performance and ruggedization are plotted, if you have no ruggedization and high performance at your data center, not surprising, you see NVIDIA, Dell, Supermicro there. If you bring down the processing and highly ruggedized stuff, you may have heard of companies like Mercury and Curtis-Wright.
They're focused in the highly rugged space. We bring the power and the benefit of both of those to the market, the highest level of performance at the highest levels of ruggedization. Where we compete today, there really isn't someone delivering similar products, so we tend to get the majority of our awards sole source. When we do compete, we're usually competing against older technology, and customers looking at making an architecture switch decision. We've been winning about 75% of those competitions over the last two years. Why that's important, the enduring market opportunity that's driving us here. If you pay attention to just this graphic on the bottom right-hand side, embedded VPX, the yellow line, represents a simplistic version of the architectures that are used in this space today for rugged edge computing.
Embedded VPX is something about the size of your iPad with a high-end Intel CPU on there, something that you might see in this laptop or your desktop at home. When that technology was invented, it hadn't really imagined AI/ML. It was more about open systems architectures, and anybody's hardware and anybody's software creates flexibility, commodity, lower prices. As you have seen and read in the press today, the oncoming movement of AI/ML demand on a platform is huge. The embedded VPX technology, that yellow line, you can only run about one to two AI algorithms or AI inferences or programs on a single card. You can imagine on an autonomous vehicle or a sensor system aircraft in the military where they have multiple sensors, radar, imagery, infrared, there's AI algorithms specifically built for each one of those systems.
You want to fuse those systems as usually done by AI. Now in a space where you were doing no, none, or one AI algorithm, now you want to do 10-12. The systems, if you use VPX, for example, struggle to meet that. We were helping a prime customer bid a combat system for the U.S. Army for a helicopter. They had envisioned putting an F-35 combat system in this helicopter, equivalent of about 52 AI inferences. That would equate to about 27 of these VPX cards, about the size of an iPad in a box. That box would be about this long and about this high. That box was almost too big to fit in the aircraft. The power of the GPU, the green line, that's what our company offers, bringing the data center to the class.
We can do 25 to 30 AI inferences in that same space, generally at 1,000 times faster latency over the Ethernet standard that the embedded VPX technology uses. What that means is that box is about six fee t long, now shrunk down to something about the size of a shoebox that was performing all of those AI inferences at 1,000 times faster latency. Not only better capability, better processing, more power at the edge, but now you significantly reduce the size, weight, and power, even though the GPU itself is higher power than a CPU, but you've reduced the number. So excited. You reduce the number of elements that do that. One thing I'll note on here is that the Department of State, about 10 months ago, started banning the sale of data center GPUs and technology to our adversaries.
The reason being is our adversaries were moving to the green line to augment their armed forces, creating an unfair advantage against our U.S. forces. You can imagine ours are embedded on the yellow line right now. The nice thing about our market is not really a question of if, but it's a question of when and how fast our DOD starts to transition over to the green line technologies and architectures, because that's what they'll need to do to create the unfair advantage. Department of State is helping by banning the sale of that technology to adversaries or countries that are contingent and sell. Company's working off about an unfactored five-year pipeline of about $1 billion. It breaks out this way across defense and commercial markets.
You'll see on the defense side, C5ISR is a fancy term for intelligence, surveillance, and reconnaissance, command and control systems in the services. Across air, ground, and maritime, that's where we're finding our applications and interest right now. On the commercial markets, we've been showing up mostly in commercial aerospace areas around autonomy that I talked about before and this composable infrastructure, this area of the data center where high density of GPUs becomes important. Our Bressner segment, interesting part of the business, it's more value-added reselling, taking the likes of Asus, Digi, NVIDIA, and integrating them and delivering to customer sets. It's been a decent-sized company for us, grows around 7%-9% a year. They've been doing on the order of 22%-24% gross margins.
As a counter to that, the OSS part of the business that's doing more of this bespoke edge computing elements has been we do generally tend to do more like 35%-45% gross margins across the business. Consolidated-wise, we're sitting in the low to mid-30s. We have been using the Bressner segment, though, as a way to facilitate sales into Europe and Germany as a channel to market to help move some of these higher-end, higher profitable, highly profitable elements in and across Europe. Where we're looking to grow, I mentioned that big pipelines. We have a five-year unfactored pipeline of about $1 billion. If we factor that down, we've generally got enough room in the OSS segment of the business to grow 20% a year domestic US organically.
We're looking to expose and grow those gross margins and use that then to generate cash and move back into the business. How this looks for us and where our focuses are today, we lead with those best-in-class rugged products. They are still built on open standards, so the Defense Department is still very interested in those because they're not bespoke specifically to one company's proprietary solution. However, the way we ruggedize them, the work we've done in the firmware and the way we integrate and build the PC board technologies that tie all these systems together is kind of like Coke secret sauce for us. There's ways we do that. The data centers are meant to run something 24/7 nonstop and don't change out on the edge, right? You're always turning things off, turning things on. You're always changing software. You're always moving data in different directions.
We had to adjust and adopt things to make that happen. We've also started to, while we do invest our money in developing standard product, we have started to build up a portfolio of customer-funded development and revenue. This is so that we can continue to establish product lines, especially on the defense side. This allows us to find an incumbent position on a platform. Imagine something like being on an aircraft or a tank or a surface ship. You can be as part of that program as an incumbent piece of the architecture for years. This allows you to get a multi-year program so that you have guaranteed steady revenue over the years. The basis of that provides steady revenue and input.
That next bullet was about, is rather than worrying on quarter to quarter, just a normal sales cycle, is by establishing platform incumbent positions, we can get these multi-year contracts. I'll show you an additional slide on why that's important. We talked about the product line already in terms of servers and storage. Some of the key things in the technology we're known for, in addition to the ones I mentioned, that latency is really driven through this standard called PCI Express. We have been a long-time leader in that technology through the six generations that it's been around and have been noted in the market as a key purveyor of that technology. I mentioned the NVIDIA areas down there below.
This just takes a look, as I mentioned, how our pipeline is kind of built out in the areas which we've kind of already talked to across the ground, air, and maritime domains in defense. One of the interesting ones I'll highlight here on the right is that 62% of our pipeline today represents opportunities that are multi-year with customers that will allow us to get to those recurring multi-year contracts. Why that's important, I would just highlight here on the right. In the overall pipeline, we have about a little better than $450 million of platform opportunities. The interesting thing about these is you have a program, you win upfront, you get a couple million dollars of development, then you move into low-rate and full-rate production, then you move into a long chain of service support and tech refresh update.
When you can establish yourself on platforms and each life cycle, especially in the Department of Defense, could last 10, 15, 20 years, you start stacking up stable year-over-year revenue. That provides for some financial stability, supports stable growth. It all starts with the customer-funded upfront development, which we've been steadily increasing since I joined the company in 2023 and see that growing significantly over the years. Here's an example of that. We won a program back in 2018, small development contract for about $1 million for a storage system for radar processing and AI on the Navy P-8 aircraft. That program since 2018, that small development program, has delivered $40 million in revenue over that seven years. We just signed this year another eight-year extension on this program.
We will be on this platform for 16 years, steadily generating high-margin revenue for us. What this looks like in the commercial world, we've got some composable infrastructure market I talked about. The multi-year contracts delivering products into these data centers, we've got some pipeline opportunity in and around $200 million to expand out over that market. Something that's probably interesting is a little more transformational for the company. We recently did a system for the U.S. Army where they were struggling to move video around a combat vehicle to the crew stations inside. They weren't able to drive the vehicle or target with the weapon system because the latency was so bad. They were using that yellow line technology. We were able to come on board and solve not only their processing capability, but the latency capability.
That program is now under field testing by the U.S. Army. While not budgeted or set yet, the U.S. Army does not buy things in tens or hundreds. They buy things in thousands. For something like us, buying systems on the order of 1,000, 2,000, or 5,000 represent three- to four- to five-year programs that are well in excess of $100 million- $200 million. We have factored these accordingly, but there are transformational opportunities like this that exist in our pipeline because of the adoption of these technologies to either update existing vehicles to a greater capability or to set the architectures for new vehicles. That is actually a pretty good summary. I am actually going to stop there and leave some time for questions. If you wanted to get to any of the topics I talked, I also have the CFO here to keep me straight on the numbers.
Feel free to ask a number of questions. We'll share what we can with you. We will release Q1 numbers here on May 6th or 7th, May 7th, so you'll be able to follow up on the continued story of the progress we've been making. Yes. [audio distortion]Yeah, great question on the supply chain, especially today and around tariffs. We design all our own PCB boards and everything that go into the system. We have a wide range of suppliers in the U.S. and outside that we can roll through that development. We do all of our design, integration, testing, and delivery out of the U.S. We will source components. If the PCB board manufacturers can't, we'll help them source components. Generally, a lot of the processing chips and all will come out of the Taiwan area. Most of our products are designed to be agnostic to GPU.
If it's not NVIDIA, it's somebody else. We watch Intel, we watch AMD, we watch others as they're developing Qualcomm. We actually did some work for the U.S. Army and the Army Special Forces on how to utilize other GPUs that might not come out of Asia so that we could protect the supply chain. Our designs are agnostic to that. We have the agility to move GPUs and other processing chipsets in and out based on that. Generally, we've been able to pass any new tariffs on to customers. We've been doing that long before the tariff craziness that's been going on over the last couple of weeks or months. We've been able to pass those on. We haven't really been extremely affected by that. In some respects, it's actually creating some opportunity for us in the data center market.
We were competing with a lot of companies out of Asia and Taiwan that had better, more competitive pricing. That has been more equaled out now. We are starting to see some more interest in the data center market on products that can be sourced and developed, delivered from the U.S.. [audiodistortion] Interestingly enough, as I mentioned, remember the way that the big prime contractors are solving those problems today are with that embedded VPX technology on that yellow line. They were not able to solve this. As soon as the Army lab, we proved to them we could, and they started doing it. It is not surprising that we are in engaged discussions now with a lot of large prime defense integrators who then would look to utilize the technology to move it forward. We are generally at that level.
We usually work through a lot of primes as an integrating element of their overall architecture on the vehicle so that we can provide and usually land and expand. We might start with a storage system. We might start with a system like this. We can work our way into the broader architecture as they want to implement more capability to bring artificial intelligence, machine learning, sensor capabilities onto the system. We have about a minute if somebody wants to squeeze one in. All right, everyone. Exciting company. We're in booth 708 over in the champagne room. No, it's not some sultry place, but being in the champagne room. Come by and talk to us. We'll be glad to do it. This presentation's up on the website. Happy to brief anybody more on the company. Look forward to you and being investors. Thanks.