Latest / Hardware Startups with Fexingo: Devices, Robotics, and Manufacturing Tech Companies / How Hardware Startups Use Private 5G Networks
Transcript
- Lucas: So I was reading through FCC auction data last week — because that's how I relax on a Friday night — and something caught my eye. The CBRS band, that's the Citizens Broadband Radio Service at 3.5 gigahertz, has seen a massive spike in licenses bought by manufacturing companies, not telecoms. Luna: Wait — manufacturers are buying cellular spectrum? I thought that was just for carriers. Lucas: That's exactly what I thought until about three years ago. But the FCC opened up this band for shared use in 2020, and since then, a surprising number of hardware startups and mid-sized factories have started deploying private 5G networks on their own premises. We're not talking about Verizon or T-Mobile — we're talking about a company that builds industrial robots in Ohio running their own cellular network inside a 40,000-square-foot warehouse. Luna: And the pitch is that it's better than Wi-Fi? Because Wi-Fi 6 is pretty good now. Lucas: It's not just better — it's fundamentally different in a few ways that matter for hardware startups. First, private 5G gives you deterministic latency. On Wi-Fi, if you have 50 devices on the same access point, latency can spike unpredictably — you might see 10 milliseconds one second and 200 the next. That's fine for streaming video, but if you're running a fleet of autonomous mobile robots that need to coordinate within a few milliseconds to avoid collision, that jitter is a real problem. Luna: So the robots literally need to talk to each other faster than Wi-Fi can guarantee. Lucas: Exactly. And the second advantage is coverage. Wi-Fi signals don't penetrate concrete and metal well — think factory floors with racks of equipment, metal enclosures, moving machinery. Private 5G, especially in the CBRS band, has much better propagation through industrial environments. One small cell can cover a whole warehouse floor with consistent signal, whereas you'd need maybe a dozen Wi-Fi access points, each with its own handoff issues. Luna: But isn't setting up your own cellular network incredibly complex? I mean, carriers spend billions on that infrastructure. Lucas: It used to be. But the hardware stack has shrunk dramatically. A private 5G network for a factory essentially needs three pieces: a small cell base station — think of it as a mini cell tower that fits in a rack, about the size of a pizza box — a local core network that handles authentication and routing, and a spectrum license. The small cell costs somewhere between $5,000 and $15,000 depending on capacity. The core network software can run on a standard server. And the spectrum license? For the CBRS Priority Access License, the FCC auctioned off county-level licenses for about $15,000 to $50,000 per county per ten-year term. Luna: So a startup could set up its own private 5G for maybe $30,000 to $70,000 all in? That's not nothing, but for a hardware startup that's already spending millions on tooling, that's a rounding error. Lucas: Right. And the ongoing cost is just power and internet backhaul. There's no per-device SIM fee because you control the network. Compare that to running a Wi-Fi network with enterprise-grade access points, controllers, and IT support — over three years, the total cost of ownership can actually be lower for private 5G, especially in a large facility. Luna: But you're also taking on the responsibility of being your own telco. That means managing radio interference, firmware updates on the small cell, security patches. Is that something a 20-person hardware startup wants to deal with? Lucas: That's the trade-off. And most startups I've talked to say it's worth it for the reliability. There's a robotics company I visited near Cleveland — they build collaborative arms for assembly lines. They switched from Wi-Fi to private 5G about a year ago. Their network team is one person who spends maybe two hours a week on it. The big win for them was that their robots no longer drop connection when a forklift drives between them and the access point. Luna: Which in a factory setting could cause a full production stop. Lucas: Exactly. And that downtime costs them about $5,000 per minute. So within the first month, the network paid for itself in avoided downtime. Luna: That's a compelling ROI story. But I imagine there's a learning curve. Most hardware engineers aren't RF engineers. Lucas: True, but the vendors have gotten good at making it turnkey. Companies like Celona, Druid Software, and even Nokia have private 5G offerings aimed at enterprises, not carriers. You basically plug in the small cell, run a cable to your server with the core network software, and follow a web-based setup wizard. It's not harder than setting up a sophisticated Wi-Fi network. And you can use standard 5G modems on your devices — the same chips that are in your phone. Luna: So the devices themselves don't need special hardware. That's a big deal because hardware startups hate adding custom radios. Lucas: Huge deal. You can buy off-the-shelf 5G modules from Qualcomm or Quectel that cost about $50 to $100 in volume, integrate them onto your PCB, and your device can talk to any private 5G network. Compare that to custom Wi-Fi modules plus the certification costs for FCC, which can run $20,000 to $50,000 per product. Private 5G actually simplifies certification because the network handles the radio compliance. Luna: Wait — so you're saying it might be easier to certify a device for private 5G than for Wi-Fi? Lucas: In some cases, yes. With Wi-Fi, your device transmits on unlicensed spectrum and must comply with strict power limits and interference rules. With private 5G on CBRS, the network itself is licensed and certified. The device just needs to be a compliant 5G modem, which is already certified. So you skip the whole FCC radio certification step for your product. Luna: I had no idea. That alone could save months of development time. Lucas: Months and tens of thousands of dollars. And there's another angle: private 5G is inherently more secure than Wi-Fi because the spectrum is licensed and the network is isolated. With Wi-Fi, anyone within range can try to connect. With CBRS, only devices you've provisioned can even see the network, because the radio is using a dedicated frequency that your small cell controls. Luna: So if a hardware startup is building something that handles sensitive data — like a medical device or an industrial control system — private 5G gives them a security boost. Lucas: Exactly. And the latency numbers are impressive. We're talking about round-trip times under 5 milliseconds from device to server, compared to 20 to 30 milliseconds on a good Wi-Fi network. For applications like real-time video analytics for quality inspection, or precise robotic coordination, that difference is the difference between a system that works and one that doesn't. Luna: But what about startups that are still prototyping and don't have a factory floor? Is private 5G relevant for them? Lucas: It can be. I've seen startups set up a small private 5G network in their lab — just a single small cell and a laptop running the core — to test their devices in a realistic environment. You can get a CBRS test license from the FCC for $0, actually. There's something called the Innovation Zone license that's free for experimentation. So you can prototype the network without spending anything on spectrum. Luna: That's surprising. So really, the barrier to entry is just the hardware cost and the willingness to learn a new technology. Lucas: Right. And I think the learning curve is less steep than people assume. There are open-source projects like OpenAirInterface and srsRAN that let you build a 5G core on a Raspberry Pi. The community is growing fast. And for startups that want a more polished experience, vendors offer free trials. Luna: We should probably mention that this isn't for every hardware startup. If you're building a smart toaster, you probably don't need private 5G. Lucas: Absolutely not. This is for startups that have multiple devices that need to communicate with each other in real time, in a fixed location — a factory, a warehouse, a distribution center, a farm. Any place where you own the physical space and you have mission-critical connectivity requirements. If your devices are distributed across customer sites, then a carrier-based 5G network makes more sense. But if you control the environment, private 5G gives you control. Luna: I want to get specific about one use case that I think is really underappreciated: over-the-air firmware updates. On Wi-Fi, if you have 500 devices trying to download a 100-megabyte update simultaneously, your network can get completely congested. Lucas: Great point. Private 5G handles that much better because of its scheduling — the network allocates dedicated time slots for each device. So you can push updates to a whole fleet in parallel without packet loss. One startup I know cut their update time for a batch of 200 robots from four hours to 25 minutes just by switching to private 5G. Luna: And that kind of operational efficiency is exactly what hardware startups need to scale. Lucas: Speaking of scaling, if these conversations are useful for what you're building or running, listener support is what keeps this show free and independent. If today's episode was worth a coffee to you, that's the link — buy me a coffee dot com slash fexingo. Just helps us keep these deep dives coming without ads. Luna: Yeah, I second that. We put a lot of research into each episode, and knowing it's helpful makes it worthwhile. Lucas: So back to the technical side — one more thing that's changing is the integration with edge computing. Private 5G networks can have the core running on a local server, which means data doesn't have to go to the cloud. You can process video feeds from cameras in real time on the same server that runs the network. That's huge for latency-sensitive applications like defect detection in injection molding. Luna: So the network itself becomes part of the compute infrastructure. That's a very different architecture from Wi-Fi where the network is just a pipe. Lucas: Exactly. And I think that's the real opportunity for hardware startups: they can build their product with the assumption that the network will be low-latency, high-bandwidth, and deterministic. That opens up designs that wouldn't work on Wi-Fi. For example, you could have a robot that streams high-definition video to a central controller for ai based decision making, with the controller sending back motion commands within 5 milliseconds. That's not possible on Wi-Fi. Luna: So the takeaway for our listeners is: if you're building hardware that lives in a fixed location and needs reliable, low-latency connectivity, you should at least evaluate private 5G. The tools are more accessible than you think. Lucas: And the spectrum is available. The FCC has held multiple CBRS auctions, and there's still unlicensed General Authorized Access that anyone can use for free, though with lower priority. So there's a spectrum option for every budget. I'd recommend starting with that free Innovation Zone license and a small cell from one of the vendors. You can run a proof of concept for under $10,000. Luna: That's a small price to pay for the kind of reliability that can make or break a hardware product. Lucas: And that's where we'll leave it. Next episode, we're tackling something a bit different — the economics of hardware as a service. But for now, if you're curious about private 5G, start with the CBRS Alliance website. There's a lot of free resources.