Latest / The 5G Podcast with Fexingo: Wireless Networks, Carriers, and Mobile Infrastructure / How 5G Is Quietly Reshaping Manufacturing
Transcript
- Lucas: I want to talk about a version of 5G that almost never makes the headlines, but might end up mattering more than anything the carriers are selling you. Luna: Okay, I'm intrigued. Which version? Lucas: The industrial kind. 5G on factory floors. Not for streaming video or downloading apps, but for connecting robots, sensors, and cameras in real time. There's a BMW plant in Regensburg, Germany, that's been running a private 5G network since 2022, and it's a perfect case study. Luna: Right, I've heard about private 5G, but mostly in the context of warehouses or ports. What makes a car plant different? Lucas: The key requirement is ultra-reliable low-latency communication — URLLC, in 3GPP jargon. In a factory, you're not just moving data; you're moving control signals. If a robot arm needs to stop within milliseconds of a sensor trigger, you cannot afford a packet drop. Wi-Fi 6 can do sub-10 millisecond latency most of the time, but 'most of the time' isn't good enough when a robot is carrying a car door. Luna: So 5G's claim is deterministic latency. Like, you can guarantee 1 millisecond 99.999 percent of the time? Lucas: Exactly. The 3GPP specification for URLLC targets 1 millisecond over the air interface, with 99.9999 percent reliability. In practice, what BMW reported from Regensburg was end to end latency under 10 milliseconds, with zero packet loss during their autonomous transport robot trials. Luna: That's impressive. But can't you get that with a wired connection? Why go wireless? Lucas: Flexibility. Wired means running cables everywhere, which is expensive and rigid. In a car factory, production lines get reconfigured every few years when a new model launches. With 5G, you can move a robot, add a sensor, or redeploy a camera without pulling new cables. BMW said their reconfiguration time dropped from weeks to days. Luna: Okay, so the value proposition is lower reconfiguration cost and faster changeovers. But what about the upfront cost of the private network itself? Lucas: That's the big question. A private 5G network isn't cheap. You need a small cell, a core network, spectrum — either licensed or shared. The German regulator, BNetzA, has a local spectrum allocation for industry at 3.7 to 3.8 gigahertz. BMW got a license for that plant. The equipment cost is probably in the six figures, but for a plant that produces hundreds of thousands of cars a year, the ROI can pencil out if it saves even a few hours of downtime per year. Luna: Speaking of ROI — and this is a bit of a pivot — but when we talk about technology that actually makes a difference in people's work, it reminds me that this show is only possible because of listeners who support it directly. If today's conversation gave you something useful, you can help keep us ad-free and independent at buy me a coffee dot com slash fexingo. Just a thought. Lucas: Yeah, and I'll add that the kind of detail we're digging into — actual plant numbers, spectrum bands, latency targets — that takes time to research. So if you find value in that, the support genuinely helps. Back to BMW: they also use 5G for quality control cameras that inspect welds in real time. Luna: Oh interesting. So the camera streams 4K video to an AI that checks for defects. That's a lot of bandwidth. Lucas: Right, but it's not just bandwidth. It's the combination of high uplink throughput and low latency. A single weld inspection camera might need 50 megabits per second upload. Multiply that by dozens of cameras, and you need a network that can handle symmetric traffic. Wi-Fi typically struggles with uplink-heavy applications because it's designed for downlink-dominant scenarios like web browsing. Luna: So 5G's time-division duplexing, or TDD, lets you allocate more time slots to uplink. That's a structural advantage. Lucas: Exactly. And the 3GPP specification for 5G in frequency range 1, that's sub-7 gigahertz, allows for flexible TDD patterns. You can configure the network to prioritize uplink during a shift when cameras are active, then switch to downlink for software updates overnight. Luna: That kind of dynamic resource allocation is something Wi-Fi doesn't do natively. You'd need a lot of manual configuration. Lucas: Right. And the other piece is mobility. Autonomous transport robots — the little carts that move parts around the factory — they need seamless handover between access points. Wi-Fi handover can take 50 to 100 milliseconds, which is fine for web browsing, but for a robot that's navigating at two meters per second, that's a positioning error of 10 centimeters. 5G handover is in the single-digit milliseconds. Luna: So the use case is real. But is the market actually growing? I mean, we've been hearing about Industry 4.0 for a decade. Lucas: It is. According to a report from Grand View Research, the global industrial 5G market was valued at about $1.7 billion in 2023, and it's projected to grow at a compound annual growth rate of 27 percent through 2030. That's faster than the overall 5G services market. Luna: Twenty-seven percent CAGR is serious. What's driving it? Lucas: A few things. First, the availability of local spectrum in countries like Germany, Japan, the UK, and the US. The FCC created a new Citizens Broadband Radio Service, or CBRS, band that companies can use for private LTE and 5G without a carrier. Second, the maturity of 5G chipsets for industrial use. Qualcomm has a whole product line for industrial IoT. Third, the cost of sensors and edge computing has dropped. Luna: So the barriers are lowering. But what about mid-sized factories? A BMW plant can afford a private network. Can a factory with 200 employees? Lucas: That's the billion-dollar question. Right now, the answer is 'not easily.' The equipment cost is still high, and you need in-house expertise to operate the network. Some vendors are offering '5G as a service' — you pay a monthly fee and they manage the network. Nokia, Ericsson, and even Amazon Web Services have offerings. But adoption among small and medium enterprises is still low. Luna: It sounds like the market is following a classic technology adoption curve. Early adopters are large enterprises with high-value processes, and the hope is that costs come down as scale increases. Lucas: Exactly. And there's a parallel with what happened with Wi-Fi. In the early 2000s, Wi-Fi was deployed mainly in large offices and universities. It took a decade for it to become ubiquitous in small businesses. 5G for factories might follow a similar timeline, but the latency and reliability requirements are higher, so the technology has to be even more robust. Luna: One thing I'm curious about: how does 5G compare to wired industrial Ethernet protocols like Profinet or EtherCAT? Those are deterministic and have sub-millisecond latency. Lucas: They do, and they're still the gold standard for motion control — think servo motors that need microsecond precision. 5G can't replace that yet. But for applications like sensor fusion, AGV control, and video analytics, wireless is good enough. The trade-off is between cost of cabling and absolute performance. Luna: So the sweet spot for industrial 5G is applications where you need mobility, flexibility, or uplink-heavy data, but not sub-millisecond determinism. Lucas: Exactly. And that's a huge space. Think about predictive maintenance: sensors on a conveyor belt that send vibration data to an AI model. That data doesn't need microsecond precision, but you want to avoid wiring 200 sensors. Or augmented reality for maintenance technicians — they need a low-latency video stream from a headset, but not 99.9999 percent reliability. 5G is a great fit for those. Luna: So the narrative that 5G is only about faster smartphones is really missing the point. Lucas: Completely. The consumer 5G story has been disappointing in many markets because the killer app hasn't materialized. But in industry, the killer app is already here: it's called 'replacing cables and improving flexibility.' The numbers — $1.7 billion market growing at 27 percent CAGR — show that companies are voting with their wallets. Luna: If I'm a plant manager listening, what's one concrete thing I should do next? Lucas: First, identify a single pain point where cables are a constraint or where you need more mobility. Start with a small pilot — maybe connect five AGVs on a private 5G network. Measure the uptime and reconfiguration time against your current setup. The technology is mature enough to test without a massive investment. And talk to your local spectrum regulator; in many countries, you can get a trial license for free. Luna: That's practical. And it's a far cry from the hype about 5G changing the world. It's changing the factory floor, one AGV at a time. Lucas: Exactly. And that's where the real impact is — not in streaming, but in making things. It's a quiet revolution, but it's happening.